GemStone 3.1 problem opening debugger

I'm trying to port my apps to 3.1. Works quite well after some rehearsal. Now I encounter an error if try to open a debugger on an object log entry. Pressing the "Debug…" button and after that selecting an entry it takes quite a while (up to a minute) and a debugger appears with the message:

Norbert,

the error is indicating the gem dropped it's connection to GemTools, so there should be an error message of some sort in the gem log ... one would hope ...

Presumably the gem died "without a trace" ... There is logic that deletes the gem log on a "normal exit" so we need to see if that logic is incorrect ...


If you `tail` the gem log after logging in with GemTools, you'll be certain to see anything that might be interesting in the log.

If nothing shows up in the log, then zip all of the logs and we'll go over them with a fine tooth comb to see if there is anything subtle going on...

Dale

----- Original Message -----
| From: "Norbert Hartl" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Monday, September 17, 2012 6:17:32 AM
| Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
|
|
|
|
| Am 17.09.2012 um 15:16 schrieb Norbert Hartl < [hidden email] >:
|
|
|
| I'm trying to port my apps to 3.1. Works quite well after some
| rehearsal. Now I encounter an error if try to open a debugger on an
| object log entry. Pressing the "Debug…" button and after that
| selecting an entry it takes quite a while (up to a minute) and a
| debugger appears with the message:
|
|
|
| Network error - text follows:
| , from GciAstFailureNetwork partner has disconnected.
|
|
| The session is invalid afterwards. I looked in the logs to find
| something but couldn't see anything useful. Neither did I manage to
| file out the stack trace. Selecting the "file out" in the debugger
| just creates a file with one entry of the stack trace. Maybe I'm
| just to stupid :)
|
|
| gemstone version:
|
|
|
| GemStone/S 64 Bit
| 3.1.0 Build: 64bit-28635
| Wed Jun 20 10:13:08 2012
|
|
| pharo version:
|
|
| GemTools-1.0-beta.8.7-31x.app
| from GemStone/S 3.1.0 Web Edition for Mac OS X
|
| The link should actually be GemTools 1.0-beta.8.7 for Gemstone/S
| 3.1.0 all platforms
|
|
| Norbert
|

Dale,

sorry but I always forget that the log files for the gems are created in a different location. I produced a stack trace in the log file. It is attached to this mail. Hope it helps!

Norbert



Am 17.09.2012 um 19:17 schrieb Dale Henrichs <[hidden email]>:

Norbert,

We're taking a look at the logs ... thanks

Dale

----- Original Message -----
| From: "Norbert Hartl" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Tuesday, September 18, 2012 4:33:57 AM
| Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
|
| Dale,
|
| sorry but I always forget that the log files for the gems are created
| in a different location. I produced a stack trace in the log file.
| It is attached to this mail. Hope it helps!
|
| Norbert
|
|
|
| Am 17.09.2012 um 19:17 schrieb Dale Henrichs <[hidden email]>:
|
| > Norbert,
| >
| > the error is indicating the gem dropped it's connection to
| > GemTools, so there should be an error message of some sort in the
| > gem log ... one would hope ...
| >
| > Presumably the gem died "without a trace" ... There is logic that
| > deletes the gem log on a "normal exit" so we need to see if that
| > logic is incorrect ...
| >
| >
| > If you `tail` the gem log after logging in with GemTools, you'll be
| > certain to see anything that might be interesting in the log.
| >
| > If nothing shows up in the log, then zip all of the logs and we'll
| > go over them with a fine tooth comb to see if there is anything
| > subtle going on...
| >
| > Dale
| >
| > ----- Original Message -----
| > | From: "Norbert Hartl" <[hidden email]>
| > | To: "GemStone Seaside beta discussion"
| > | <[hidden email]>
| > | Sent: Monday, September 17, 2012 6:17:32 AM
| > | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
| > |
| > |
| > |
| > |
| > | Am 17.09.2012 um 15:16 schrieb Norbert Hartl < [hidden email]
| > | >:
| > |
| > |
| > |
| > | I'm trying to port my apps to 3.1. Works quite well after some
| > | rehearsal. Now I encounter an error if try to open a debugger on
| > | an
| > | object log entry. Pressing the "Debug…" button and after that
| > | selecting an entry it takes quite a while (up to a minute) and a
| > | debugger appears with the message:
| > |
| > |
| > |
| > | Network error - text follows:
| > | , from GciAstFailureNetwork partner has disconnected.
| > |
| > |
| > | The session is invalid afterwards. I looked in the logs to find
| > | something but couldn't see anything useful. Neither did I manage
| > | to
| > | file out the stack trace. Selecting the "file out" in the
| > | debugger
| > | just creates a file with one entry of the stack trace. Maybe I'm
| > | just to stupid :)
| > |
| > |
| > | gemstone version:
| > |
| > |
| > |
| > | GemStone/S 64 Bit
| > | 3.1.0 Build: 64bit-28635
| > | Wed Jun 20 10:13:08 2012
| > |
| > |
| > | pharo version:
| > |
| > |
| > | GemTools-1.0-beta.8.7-31x.app
| > | from GemStone/S 3.1.0 Web Edition for Mac OS X
| > |
| > | The link should actually be GemTools 1.0-beta.8.7 for Gemstone/S
| > | 3.1.0 all platforms
| > |
| > |
| > | Norbert
| > |
|
|

Norbert,

The stack traces in the log show that the error is occurring in the generated native code, so we'll need to gather additional information about the bug.

Before we go too far, we  recently released 3.1.0.1 with a number of bugfixes and there's a chance that the issue you are seeing is already addressed.

So if you could switch to doing your work in 3.1.0.1[1] the problem may not appear again. If it does we'll debug it against 3.1.0.1.

I'm just starting to write a blog post about 3.1.0.1, so I haven't updated the install scripts (I'm planning on producing a generic install script) nor have I double checked whether the 3.1 GemTools works against 3.1.0.1. Other than that there should be no differences in procedures as 3.1.0.1 is a purely bugfix release.

If you need to continue working in 3.1 for some reason, then let us know and we'll work on debugging the problem there.

Dale

[1] ftp://ftp.gemstone.com/pub/GemStone64/3.1.0.1/

----- Original Message -----
| From: "Norbert Hartl" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Tuesday, September 18, 2012 4:33:57 AM
| Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
|
| Dale,
|
| sorry but I always forget that the log files for the gems are created
| in a different location. I produced a stack trace in the log file.
| It is attached to this mail. Hope it helps!
|
| Norbert
|
|
|
| Am 17.09.2012 um 19:17 schrieb Dale Henrichs <[hidden email]>:
|
| > Norbert,
| >
| > the error is indicating the gem dropped it's connection to
| > GemTools, so there should be an error message of some sort in the
| > gem log ... one would hope ...
| >
| > Presumably the gem died "without a trace" ... There is logic that
| > deletes the gem log on a "normal exit" so we need to see if that
| > logic is incorrect ...
| >
| >
| > If you `tail` the gem log after logging in with GemTools, you'll be
| > certain to see anything that might be interesting in the log.
| >
| > If nothing shows up in the log, then zip all of the logs and we'll
| > go over them with a fine tooth comb to see if there is anything
| > subtle going on...
| >
| > Dale
| >
| > ----- Original Message -----
| > | From: "Norbert Hartl" <[hidden email]>
| > | To: "GemStone Seaside beta discussion"
| > | <[hidden email]>
| > | Sent: Monday, September 17, 2012 6:17:32 AM
| > | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
| > |
| > |
| > |
| > |
| > | Am 17.09.2012 um 15:16 schrieb Norbert Hartl < [hidden email]
| > | >:
| > |
| > |
| > |
| > | I'm trying to port my apps to 3.1. Works quite well after some
| > | rehearsal. Now I encounter an error if try to open a debugger on
| > | an
| > | object log entry. Pressing the "Debug…" button and after that
| > | selecting an entry it takes quite a while (up to a minute) and a
| > | debugger appears with the message:
| > |
| > |
| > |
| > | Network error - text follows:
| > | , from GciAstFailureNetwork partner has disconnected.
| > |
| > |
| > | The session is invalid afterwards. I looked in the logs to find
| > | something but couldn't see anything useful. Neither did I manage
| > | to
| > | file out the stack trace. Selecting the "file out" in the
| > | debugger
| > | just creates a file with one entry of the stack trace. Maybe I'm
| > | just to stupid :)
| > |
| > |
| > | gemstone version:
| > |
| > |
| > |
| > | GemStone/S 64 Bit
| > | 3.1.0 Build: 64bit-28635
| > | Wed Jun 20 10:13:08 2012
| > |
| > |
| > | pharo version:
| > |
| > |
| > | GemTools-1.0-beta.8.7-31x.app
| > | from GemStone/S 3.1.0 Web Edition for Mac OS X
| > |
| > | The link should actually be GemTools 1.0-beta.8.7 for Gemstone/S
| > | 3.1.0 all platforms
| > |
| > |
| > | Norbert
| > |
|
|

Hello,

Larry,

This bug isn't necessarily related to the debugger ... Norbert is accessing a continuation with the expectation of getting a debugger, but the SigSegv is killing the gem, so the error appears to be at the object level or the code generation level.

In general you should be able to proceed and accept methods in the debugger, however when debugging continuations from the ObjectLog, proceeding is not allowed.

I'd be interested in more information about trouble accepting methods or proceeding (when continuations aren't involved).

Dale

----- Original Message -----
| From: "Lawrence Kellogg" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Wednesday, September 19, 2012 10:58:36 AM
| Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
|
| Hello,
| Speaking of the debugger, does it work properly under GemStone 3.1? I
| am not able to proceed under 2.4.4.1, or change methods and accept
| them. Is it something I'm doing wrong?
|
|
| Regards,
|
|
| Larry Kellogg
|
|
|
|
|
| On Sep 19, 2012, at 1:25 PM, Dale Henrichs wrote:
|
|
|
| Norbert,
|
| The stack traces in the log show that the error is occurring in the
| generated native code, so we'll need to gather additional
| information about the bug.
|
| Before we go too far, we recently released 3.1.0.1 with a number of
| bugfixes and there's a chance that the issue you are seeing is
| already addressed.
|
| So if you could switch to doing your work in 3.1.0.1[1] the problem
| may not appear again. If it does we'll debug it against 3.1.0.1.
|
| I'm just starting to write a blog post about 3.1.0.1, so I haven't
| updated the install scripts (I'm planning on producing a generic
| install script) nor have I double checked whether the 3.1 GemTools
| works against 3.1.0.1. Other than that there should be no
| differences in procedures as 3.1.0.1 is a purely bugfix release.
|
| If you need to continue working in 3.1 for some reason, then let us
| know and we'll work on debugging the problem there.
|
| Dale
|
| [1] ftp://ftp.gemstone.com/pub/GemStone64/3.1.0.1/
|
| ----- Original Message -----
| | From: "Norbert Hartl" < [hidden email] >
| | To: "GemStone Seaside beta discussion" < [hidden email]
| | >
| | Sent: Tuesday, September 18, 2012 4:33:57 AM
| | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
| |
| | Dale,
| |
| | sorry but I always forget that the log files for the gems are
| | created
| | in a different location. I produced a stack trace in the log file.
| | It is attached to this mail. Hope it helps!
| |
| | Norbert
| |
| |
| |
| | Am 17.09.2012 um 19:17 schrieb Dale Henrichs < [hidden email]
| | >:
| |
| | > Norbert,
| | >
| | > the error is indicating the gem dropped it's connection to
| | > GemTools, so there should be an error message of some sort in the
| | > gem log ... one would hope ...
| | >
| | > Presumably the gem died "without a trace" ... There is logic that
| | > deletes the gem log on a "normal exit" so we need to see if that
| | > logic is incorrect ...
| | >
| | >
| | > If you `tail` the gem log after logging in with GemTools, you'll
| | > be
| | > certain to see anything that might be interesting in the log.
| | >
| | > If nothing shows up in the log, then zip all of the logs and
| | > we'll
| | > go over them with a fine tooth comb to see if there is anything
| | > subtle going on...
| | >
| | > Dale
| | >
| | > ----- Original Message -----
| | > | From: "Norbert Hartl" < [hidden email] >
| | > | To: "GemStone Seaside beta discussion"
| | > | < [hidden email] >
| | > | Sent: Monday, September 17, 2012 6:17:32 AM
| | > | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
| | > |
| | > |
| | > |
| | > |
| | > | Am 17.09.2012 um 15:16 schrieb Norbert Hartl <
| | > | [hidden email]
| | > | >:
| | > |
| | > |
| | > |
| | > | I'm trying to port my apps to 3.1. Works quite well after some
| | > | rehearsal. Now I encounter an error if try to open a debugger
| | > | on
| | > | an
| | > | object log entry. Pressing the "Debug…" button and after that
| | > | selecting an entry it takes quite a while (up to a minute) and
| | > | a
| | > | debugger appears with the message:
| | > |
| | > |
| | > |
| | > | Network error - text follows:
| | > | , from GciAstFailureNetwork partner has disconnected.
| | > |
| | > |
| | > | The session is invalid afterwards. I looked in the logs to find
| | > | something but couldn't see anything useful. Neither did I
| | > | manage
| | > | to
| | > | file out the stack trace. Selecting the "file out" in the
| | > | debugger
| | > | just creates a file with one entry of the stack trace. Maybe
| | > | I'm
| | > | just to stupid :)
| | > |
| | > |
| | > | gemstone version:
| | > |
| | > |
| | > |
| | > | GemStone/S 64 Bit
| | > | 3.1.0 Build: 64bit-28635
| | > | Wed Jun 20 10:13:08 2012
| | > |
| | > |
| | > | pharo version:
| | > |
| | > |
| | > | GemTools-1.0-beta.8.7-31x.app
| | > | from GemStone/S 3.1.0 Web Edition for Mac OS X
| | > |
| | > | The link should actually be GemTools 1.0-beta.8.7 for
| | > | Gemstone/S
| | > | 3.1.0 all platforms
| | > |
| | > |
| | > | Norbert
| | > |
| |
| |
|
|

On Sep 19, 2012, at 2:53 PM, Dale Henrichs wrote:

> Larry,
>
> This bug isn't necessarily related to the debugger ... Norbert is accessing a continuation with the expectation of getting a debugger, but the SigSegv is killing the gem, so the error appears to be at the object level or the code generation level.
>
> In general you should be able to proceed and accept methods in the debugger, however when debugging continuations from the ObjectLog, proceeding is not allowed.
>
> I'd be interested in more information about trouble accepting methods or proceeding (when continuations aren't involved).
>

  Well, I'm feeling chagrined. I wrote my whole app using a local Pharo image and when I moved it to GemStone, the walkbacks I got were continuations from the ObjectLog, which I could not accept or continue from, so I thought all of the debugging was broken. Silly me! It works fine if I put a 'self halt' in a method. I can't believe I never tried that on my server. Funny that I can write server code without needing the debugger. My methods work most of the time on the first try.

  Thanks Dale!

Larry

Larry,

No Problem ... there's always room for learning and I am glad that you are comfortable asking the questions!

Dale

----- Original Message -----
| From: "Lawrence Kellogg" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Wednesday, September 19, 2012 12:16:38 PM
| Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
|
|
| On Sep 19, 2012, at 2:53 PM, Dale Henrichs wrote:
|
| > Larry,
| >
| > This bug isn't necessarily related to the debugger ... Norbert is
| > accessing a continuation with the expectation of getting a
| > debugger, but the SigSegv is killing the gem, so the error appears
| > to be at the object level or the code generation level.
| >
| > In general you should be able to proceed and accept methods in the
| > debugger, however when debugging continuations from the ObjectLog,
| > proceeding is not allowed.
| >
| > I'd be interested in more information about trouble accepting
| > methods or proceeding (when continuations aren't involved).
| >
|
|   Well, I'm feeling chagrined. I wrote my whole app using a local
|   Pharo image and when I moved it to GemStone, the walkbacks I got
|   were continuations from the ObjectLog, which I could not accept or
|   continue from, so I thought all of the debugging was broken. Silly
|   me! It works fine if I put a 'self halt' in a method. I can't
|   believe I never tried that on my server. Funny that I can write
|   server code without needing the debugger. My methods work most of
|   the time on the first try.
|
|   Thanks Dale!
|
| Larry
|
|
|
| > Dale
| >
| > ----- Original Message -----
| > | From: "Lawrence Kellogg" <[hidden email]>
| > | To: "GemStone Seaside beta discussion"
| > | <[hidden email]>
| > | Sent: Wednesday, September 19, 2012 10:58:36 AM
| > | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
| > |
| > | Hello,
| > | Speaking of the debugger, does it work properly under GemStone
| > | 3.1? I
| > | am not able to proceed under 2.4.4.1, or change methods and
| > | accept
| > | them. Is it something I'm doing wrong?
| > |
| > |
| > | Regards,
| > |
| > |
| > | Larry Kellogg
| > |
| > |
| > |
| > |
| > |
| > | On Sep 19, 2012, at 1:25 PM, Dale Henrichs wrote:
| > |
| > |
| > |
| > | Norbert,
| > |
| > | The stack traces in the log show that the error is occurring in
| > | the
| > | generated native code, so we'll need to gather additional
| > | information about the bug.
| > |
| > | Before we go too far, we recently released 3.1.0.1 with a number
| > | of
| > | bugfixes and there's a chance that the issue you are seeing is
| > | already addressed.
| > |
| > | So if you could switch to doing your work in 3.1.0.1[1] the
| > | problem
| > | may not appear again. If it does we'll debug it against 3.1.0.1.
| > |
| > | I'm just starting to write a blog post about 3.1.0.1, so I
| > | haven't
| > | updated the install scripts (I'm planning on producing a generic
| > | install script) nor have I double checked whether the 3.1
| > | GemTools
| > | works against 3.1.0.1. Other than that there should be no
| > | differences in procedures as 3.1.0.1 is a purely bugfix release.
| > |
| > | If you need to continue working in 3.1 for some reason, then let
| > | us
| > | know and we'll work on debugging the problem there.
| > |
| > | Dale
| > |
| > | [1] ftp://ftp.gemstone.com/pub/GemStone64/3.1.0.1/
| > |
| > | ----- Original Message -----
| > | | From: "Norbert Hartl" < [hidden email] >
| > | | To: "GemStone Seaside beta discussion" <
| > | | [hidden email]
| > | | >
| > | | Sent: Tuesday, September 18, 2012 4:33:57 AM
| > | | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
| > | |
| > | | Dale,
| > | |
| > | | sorry but I always forget that the log files for the gems are
| > | | created
| > | | in a different location. I produced a stack trace in the log
| > | | file.
| > | | It is attached to this mail. Hope it helps!
| > | |
| > | | Norbert
| > | |
| > | |
| > | |
| > | | Am 17.09.2012 um 19:17 schrieb Dale Henrichs <
| > | | [hidden email]
| > | | >:
| > | |
| > | | > Norbert,
| > | | >
| > | | > the error is indicating the gem dropped it's connection to
| > | | > GemTools, so there should be an error message of some sort in
| > | | > the
| > | | > gem log ... one would hope ...
| > | | >
| > | | > Presumably the gem died "without a trace" ... There is logic
| > | | > that
| > | | > deletes the gem log on a "normal exit" so we need to see if
| > | | > that
| > | | > logic is incorrect ...
| > | | >
| > | | >
| > | | > If you `tail` the gem log after logging in with GemTools,
| > | | > you'll
| > | | > be
| > | | > certain to see anything that might be interesting in the log.
| > | | >
| > | | > If nothing shows up in the log, then zip all of the logs and
| > | | > we'll
| > | | > go over them with a fine tooth comb to see if there is
| > | | > anything
| > | | > subtle going on...
| > | | >
| > | | > Dale
| > | | >
| > | | > ----- Original Message -----
| > | | > | From: "Norbert Hartl" < [hidden email] >
| > | | > | To: "GemStone Seaside beta discussion"
| > | | > | < [hidden email] >
| > | | > | Sent: Monday, September 17, 2012 6:17:32 AM
| > | | > | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening
| > | | > | debugger
| > | | > |
| > | | > |
| > | | > |
| > | | > |
| > | | > | Am 17.09.2012 um 15:16 schrieb Norbert Hartl <
| > | | > | [hidden email]
| > | | > | >:
| > | | > |
| > | | > |
| > | | > |
| > | | > | I'm trying to port my apps to 3.1. Works quite well after
| > | | > | some
| > | | > | rehearsal. Now I encounter an error if try to open a
| > | | > | debugger
| > | | > | on
| > | | > | an
| > | | > | object log entry. Pressing the "Debug…" button and after
| > | | > | that
| > | | > | selecting an entry it takes quite a while (up to a minute)
| > | | > | and
| > | | > | a
| > | | > | debugger appears with the message:
| > | | > |
| > | | > |
| > | | > |
| > | | > | Network error - text follows:
| > | | > | , from GciAstFailureNetwork partner has disconnected.
| > | | > |
| > | | > |
| > | | > | The session is invalid afterwards. I looked in the logs to
| > | | > | find
| > | | > | something but couldn't see anything useful. Neither did I
| > | | > | manage
| > | | > | to
| > | | > | file out the stack trace. Selecting the "file out" in the
| > | | > | debugger
| > | | > | just creates a file with one entry of the stack trace.
| > | | > | Maybe
| > | | > | I'm
| > | | > | just to stupid :)
| > | | > |
| > | | > |
| > | | > | gemstone version:
| > | | > |
| > | | > |
| > | | > |
| > | | > | GemStone/S 64 Bit
| > | | > | 3.1.0 Build: 64bit-28635
| > | | > | Wed Jun 20 10:13:08 2012
| > | | > |
| > | | > |
| > | | > | pharo version:
| > | | > |
| > | | > |
| > | | > | GemTools-1.0-beta.8.7-31x.app
| > | | > | from GemStone/S 3.1.0 Web Edition for Mac OS X
| > | | > |
| > | | > | The link should actually be GemTools 1.0-beta.8.7 for
| > | | > | Gemstone/S
| > | | > | 3.1.0 all platforms
| > | | > |
| > | | > |
| > | | > | Norbert
| > | | > |
| > | |
| > | |
| > |
| > |
|
|

Dale,

Am 19.09.2012 um 19:25 schrieb Dale Henrichs <[hidden email]>:
I updated my installation to 3.1.0.1 but get the same result. What I did:

- changed the $vers variable in the install script from 3.1.0 to 3.1.0.1
- installed gemstone with the install script
- starting my current extent with the new gemstone version
- using the one click image brings up a dialog that it is incompatible with the server version
- copied the gci and ssl lib from the new gemstone installation into the one click resources directory
- connecting again brings up the same thing as in my first mail

then I copied the extent to my local machine (Mac OS instead of Linux) and started it there. If I select a continuation I immediately get temporary object memory is full. My gems have 50MB temporary object memory.

Before I continue debugging on my own it might be good if you tell me what I should do next.

Norbert

Norbert,

If you want to look at the 3.1 extent from 3.1.0.1, then you need to upgrade the extent first.

The upgrade instructions are in the installation guide.

It is probably safe to skip the upgradeSeasideImage step when upgrading between 3.1 and 3.1.01, but I haven't tried it (yet). As I said I'm working on the 3.1.0.1 blog post so I will give that a try myself before publishing...

You should have gotten an error during stone startup (I'd think) when trying to use an extent from a different version...it's worth taking a look at the 3.1 stone log to see if you got a warning there...

Dale

----- Original Message -----
| From: "Norbert Hartl" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Thursday, September 20, 2012 5:00:20 AM
| Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
|
| Dale,
|
| Am 19.09.2012 um 19:25 schrieb Dale Henrichs <[hidden email]>:
|
| > Norbert,
| >
| > The stack traces in the log show that the error is occurring in the
| > generated native code, so we'll need to gather additional
| > information about the bug.
| >
| > Before we go too far, we  recently released 3.1.0.1 with a number
| > of bugfixes and there's a chance that the issue you are seeing is
| > already addressed.
| >
| > So if you could switch to doing your work in 3.1.0.1[1] the problem
| > may not appear again. If it does we'll debug it against 3.1.0.1.
| >
| > I'm just starting to write a blog post about 3.1.0.1, so I haven't
| > updated the install scripts (I'm planning on producing a generic
| > install script) nor have I double checked whether the 3.1 GemTools
| > works against 3.1.0.1. Other than that there should be no
| > differences in procedures as 3.1.0.1 is a purely bugfix release.
| >
| > If you need to continue working in 3.1 for some reason, then let us
| > know and we'll work on debugging the problem there.
| >
| I updated my installation to 3.1.0.1 but get the same result. What I
| did:
|
| - changed the $vers variable in the install script from 3.1.0 to
| 3.1.0.1
| - installed gemstone with the install script
| - starting my current extent with the new gemstone version
| - using the one click image brings up a dialog that it is
| incompatible with the server version
| - copied the gci and ssl lib from the new gemstone installation into
| the one click resources directory
| - connecting again brings up the same thing as in my first mail
|
| then I copied the extent to my local machine (Mac OS instead of
| Linux) and started it there. If I select a continuation I
| immediately get temporary object memory is full. My gems have 50MB
| temporary object memory.
|
| Before I continue debugging on my own it might be good if you tell me
| what I should do next.
|
| Norbert
|
|
| > Dale
| >
| > [1] ftp://ftp.gemstone.com/pub/GemStone64/3.1.0.1/
| >
| > ----- Original Message -----
| > | From: "Norbert Hartl" <[hidden email]>
| > | To: "GemStone Seaside beta discussion"
| > | <[hidden email]>
| > | Sent: Tuesday, September 18, 2012 4:33:57 AM
| > | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
| > |
| > | Dale,
| > |
| > | sorry but I always forget that the log files for the gems are
| > | created
| > | in a different location. I produced a stack trace in the log
| > | file.
| > | It is attached to this mail. Hope it helps!
| > |
| > | Norbert
| > |
| > |
| > |
| > | Am 17.09.2012 um 19:17 schrieb Dale Henrichs
| > | <[hidden email]>:
| > |
| > | > Norbert,
| > | >
| > | > the error is indicating the gem dropped it's connection to
| > | > GemTools, so there should be an error message of some sort in
| > | > the
| > | > gem log ... one would hope ...
| > | >
| > | > Presumably the gem died "without a trace" ... There is logic
| > | > that
| > | > deletes the gem log on a "normal exit" so we need to see if
| > | > that
| > | > logic is incorrect ...
| > | >
| > | >
| > | > If you `tail` the gem log after logging in with GemTools,
| > | > you'll be
| > | > certain to see anything that might be interesting in the log.
| > | >
| > | > If nothing shows up in the log, then zip all of the logs and
| > | > we'll
| > | > go over them with a fine tooth comb to see if there is anything
| > | > subtle going on...
| > | >
| > | > Dale
| > | >
| > | > ----- Original Message -----
| > | > | From: "Norbert Hartl" <[hidden email]>
| > | > | To: "GemStone Seaside beta discussion"
| > | > | <[hidden email]>
| > | > | Sent: Monday, September 17, 2012 6:17:32 AM
| > | > | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening
| > | > | debugger
| > | > |
| > | > |
| > | > |
| > | > |
| > | > | Am 17.09.2012 um 15:16 schrieb Norbert Hartl <
| > | > | [hidden email]
| > | > | >:
| > | > |
| > | > |
| > | > |
| > | > | I'm trying to port my apps to 3.1. Works quite well after
| > | > | some
| > | > | rehearsal. Now I encounter an error if try to open a debugger
| > | > | on
| > | > | an
| > | > | object log entry. Pressing the "Debug…" button and after that
| > | > | selecting an entry it takes quite a while (up to a minute)
| > | > | and a
| > | > | debugger appears with the message:
| > | > |
| > | > |
| > | > |
| > | > | Network error - text follows:
| > | > | , from GciAstFailureNetwork partner has disconnected.
| > | > |
| > | > |
| > | > | The session is invalid afterwards. I looked in the logs to
| > | > | find
| > | > | something but couldn't see anything useful. Neither did I
| > | > | manage
| > | > | to
| > | > | file out the stack trace. Selecting the "file out" in the
| > | > | debugger
| > | > | just creates a file with one entry of the stack trace. Maybe
| > | > | I'm
| > | > | just to stupid :)
| > | > |
| > | > |
| > | > | gemstone version:
| > | > |
| > | > |
| > | > |
| > | > | GemStone/S 64 Bit
| > | > | 3.1.0 Build: 64bit-28635
| > | > | Wed Jun 20 10:13:08 2012
| > | > |
| > | > |
| > | > | pharo version:
| > | > |
| > | > |
| > | > | GemTools-1.0-beta.8.7-31x.app
| > | > | from GemStone/S 3.1.0 Web Edition for Mac OS X
| > | > |
| > | > | The link should actually be GemTools 1.0-beta.8.7 for
| > | > | Gemstone/S
| > | > | 3.1.0 all platforms
| > | > |
| > | > |
| > | > | Norbert
| > | > |
| > |
| > |
|
|

Dale,

Am 20.09.2012 um 17:12 schrieb Dale Henrichs <[hidden email]>:

> Norbert,
>
> If you want to look at the 3.1 extent from 3.1.0.1, then you need to upgrade the extent first.
>
uh, oh, yes, I didn't expect something like this while upgrading to a version that has altered after the third dot. On the other hand it is totally clear that "bug fix" can also applied on the smalltalk side. :)

Norbert

Dale,


Am 20.09.2012 um 17:12 schrieb Dale Henrichs <[hidden email]>:

> Norbert,
>
> If you want to look at the 3.1 extent from 3.1.0.1, then you need to upgrade the extent first.
>
> The upgrade instructions are in the installation guide.
>
Where can I find the 3.1.0.1 guide? I only have the 3.1.0 guide.

> It is probably safe to skip the upgradeSeasideImage step when upgrading between 3.1 and 3.1.01, but I haven't tried it (yet). As I said I'm working on the 3.1.0.1 blog post so I will give that a try myself before publishing...
>
> You should have gotten an error during stone startup (I'd think) when trying to use an extent from a different version...it's worth taking a look at the 3.1 stone log to see if you got a warning there…
>
I didn't get an error when starting the 3.1.0.1 stone with an 3.1.0 extent. Find the log.txt attached.

----- Original Message -----
| From: "Norbert Hartl" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Thursday, September 20, 2012 9:04:23 AM
| Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
|
| Dale,
|
|
| Am 20.09.2012 um 17:12 schrieb Dale Henrichs <[hidden email]>:
|
| > Norbert,
| >
| > If you want to look at the 3.1 extent from 3.1.0.1, then you need
| > to upgrade the extent first.
| >
| > The upgrade instructions are in the installation guide.
| >
| Where can I find the 3.1.0.1 guide? I only have the 3.1.0 guide.

The latest docs can always be found here[1]. Older versions of the docs can be found on the sidebar...

Dale

[1] http://community.gemstone.com/display/GSS64/GemStoneS+64+Documentation

Dale,

I cannot find an 3.1.0.1 installation guide only a 3.1.0 one and I followed it. Anyway, just to make it more clear. I installed 3.1.0.1 and converted my 2.4.4.1 image using the upgrade* scripts from the new installation which updated it to 3.1.0.1. Then I remembered I forgot to upgrade the scripts I use to start a web gem. I updated those as well.
Finally I downloaded again the one-click 3.1.0.1 to be on the safe side. It is still the same problem. There is something strange going on. I can start everything fresh and do not issue a web request. When I log in using gemtools and try to open a debugger the error occurs. After that I can even see an error in the web gems log file saying

-----------  Unreportable ERROR Encountered: 2012-09-22T11:19:06.81087493896484+02:00a ImproperOperation occurred (error 2089), reason:rtErrBadCopyFromTo, An index range was specified for a sequenceable collection with the starting index < a lot of ^@ characters > ( ...) greater than the ending index 0.-----------

To this mail I attach a set of log files that contain everything from the start of the stone to the crash and side effects. Hope this helps. If not you need to guide me to test other things.

Norbert



 
Am 20.09.2012 um 21:39 schrieb Dale Henrichs <[hidden email]>:

Norbert,

When you "open the debugger" are you opening the debugger against a continuation that was created in 2.4.4.1 or are you "opening the debugger" on a new continuation ... or ?

Dale

----- Original Message -----
| From: "Norbert Hartl" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Saturday, September 22, 2012 2:28:53 AM
| Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
|
| Dale,
|
| I cannot find an 3.1.0.1 installation guide only a 3.1.0 one and I
| followed it. Anyway, just to make it more clear. I installed 3.1.0.1
| and converted my 2.4.4.1 image using the upgrade* scripts from the
| new installation which updated it to 3.1.0.1. Then I remembered I
| forgot to upgrade the scripts I use to start a web gem. I updated
| those as well.
| Finally I downloaded again the one-click 3.1.0.1 to be on the safe
| side. It is still the same problem. There is something strange going
| on. I can start everything fresh and do not issue a web request.
| When I log in using gemtools and try to open a debugger the error
| occurs. After that I can even see an error in the web gems log file
| saying
|
| -----------  Unreportable ERROR Encountered:
| 2012-09-22T11:19:06.81087493896484+02:00a ImproperOperation occurred
| (error 2089), reason:rtErrBadCopyFromTo, An index range was
| specified for a sequenceable collection with the starting index < a
| lot of ^@ characters > ( ...) greater than the ending index
| 0.-----------
|
| To this mail I attach a set of log files that contain everything from
| the start of the stone to the crash and side effects. Hope this
| helps. If not you need to guide me to test other things.
|
| Norbert
|
|
|
|  
| Am 20.09.2012 um 21:39 schrieb Dale Henrichs <[hidden email]>:
|
| >
| >
| > ----- Original Message -----
| > | From: "Norbert Hartl" <[hidden email]>
| > | To: "GemStone Seaside beta discussion"
| > | <[hidden email]>
| > | Sent: Thursday, September 20, 2012 9:04:23 AM
| > | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
| > |
| > | Dale,
| > |
| > |
| > | Am 20.09.2012 um 17:12 schrieb Dale Henrichs
| > | <[hidden email]>:
| > |
| > | > Norbert,
| > | >
| > | > If you want to look at the 3.1 extent from 3.1.0.1, then you
| > | > need
| > | > to upgrade the extent first.
| > | >
| > | > The upgrade instructions are in the installation guide.
| > | >
| > | Where can I find the 3.1.0.1 guide? I only have the 3.1.0 guide.
| >
| > The latest docs can always be found here[1]. Older versions of the
| > docs can be found on the sidebar...
| >
| > Dale
| >
| > [1]
| > http://community.gemstone.com/display/GSS64/GemStoneS+64+Documentation
| >
| >
|
|

Norbert,

It looks like you are hitting the same problem, so we'll need to debug it in more detail. At least we've ruled out the possibility that the bug is already fixed in 3.1.0.1...

We'll use gemnetdebug to get a better stack trace and you need to follow these steps to work around some bugs related to gemnetdebug:

   cd $GEMSTONE/libSlow
   chmod +w .
   cp $GEMSTONE/lib/libssl-3.1.0.1-64.so .

  Install the attached gemnetdebug in his $GEMSTONE/sys to get
  the necessary edits (need to change write permissions on
  file and set execute permissions after copy)

  Use gemnetdebug instead of gemnetobject for the login ,
  then try to open the debugger.  We need both the gemnetobj*.log
  and the gem*_code.log .

To use gemnetdebug instead of gemnetobject for a GemTools login. Edit you session in GemTools Launcher and copy the template to a workspace for reference. Next create a New Session and select the 'Custom' option. Edit the 'Custom' templates using the information from your original template (add 'debug' to the name of the session' and change the line:

  gemTask: 'gemnetobject';

to:

  gemTask: 'gemnetdebug';

then save the session. Use this session to login and open the debugger ... when you get the SIGSEGV, the stack trace should make more sense to us.

Dale
----- Original Message -----
| From: "Norbert Hartl" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Saturday, September 22, 2012 2:28:53 AM
| Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
|
| Dale,
|
| I cannot find an 3.1.0.1 installation guide only a 3.1.0 one and I
| followed it. Anyway, just to make it more clear. I installed 3.1.0.1
| and converted my 2.4.4.1 image using the upgrade* scripts from the
| new installation which updated it to 3.1.0.1. Then I remembered I
| forgot to upgrade the scripts I use to start a web gem. I updated
| those as well.
| Finally I downloaded again the one-click 3.1.0.1 to be on the safe
| side. It is still the same problem. There is something strange going
| on. I can start everything fresh and do not issue a web request.
| When I log in using gemtools and try to open a debugger the error
| occurs. After that I can even see an error in the web gems log file
| saying
|
| -----------  Unreportable ERROR Encountered:
| 2012-09-22T11:19:06.81087493896484+02:00a ImproperOperation occurred
| (error 2089), reason:rtErrBadCopyFromTo, An index range was
| specified for a sequenceable collection with the starting index < a
| lot of ^@ characters > ( ...) greater than the ending index
| 0.-----------
|
| To this mail I attach a set of log files that contain everything from
| the start of the stone to the crash and side effects. Hope this
| helps. If not you need to guide me to test other things.
|
| Norbert
|
|
|
|  
| Am 20.09.2012 um 21:39 schrieb Dale Henrichs <[hidden email]>:
|
| >
| >
| > ----- Original Message -----
| > | From: "Norbert Hartl" <[hidden email]>
| > | To: "GemStone Seaside beta discussion"
| > | <[hidden email]>
| > | Sent: Thursday, September 20, 2012 9:04:23 AM
| > | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
| > |
| > | Dale,
| > |
| > |
| > | Am 20.09.2012 um 17:12 schrieb Dale Henrichs
| > | <[hidden email]>:
| > |
| > | > Norbert,
| > | >
| > | > If you want to look at the 3.1 extent from 3.1.0.1, then you
| > | > need
| > | > to upgrade the extent first.
| > | >
| > | > The upgrade instructions are in the installation guide.
| > | >
| > | Where can I find the 3.1.0.1 guide? I only have the 3.1.0 guide.
| >
| > The latest docs can always be found here[1]. Older versions of the
| > docs can be found on the sidebar...
| >
| > Dale
| >
| > [1]
| > http://community.gemstone.com/display/GSS64/GemStoneS+64+Documentation
| >
| >
|
|

Norbert,

Regarding the Unreportable ERROR, I'm not quite sure that the two different problems are related ... At first blush the error looks similar to the error I would expect if the fastCGI port was being hit with an empty request coming from something like monit...

I am curious how you are upgrading your application. I assume you are using GLASS 1.0-beta.8.7.3 and Seaside 3.0.7.1 like I described in my blog post[1].

If you are not using those two versions in GemStone 3.1.0.1, then perhaps the Unreportable ERROR is coming from that direction...

Dale

[1] http://gemstonesoup.wordpress.com/#helper

----- Original Message -----
| From: "Norbert Hartl" <[hidden email]>
| To: "GemStone Seaside beta discussion" <[hidden email]>
| Sent: Saturday, September 22, 2012 2:28:53 AM
| Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
|
| Dale,
|
| I cannot find an 3.1.0.1 installation guide only a 3.1.0 one and I
| followed it. Anyway, just to make it more clear. I installed 3.1.0.1
| and converted my 2.4.4.1 image using the upgrade* scripts from the
| new installation which updated it to 3.1.0.1. Then I remembered I
| forgot to upgrade the scripts I use to start a web gem. I updated
| those as well.
| Finally I downloaded again the one-click 3.1.0.1 to be on the safe
| side. It is still the same problem. There is something strange going
| on. I can start everything fresh and do not issue a web request.
| When I log in using gemtools and try to open a debugger the error
| occurs. After that I can even see an error in the web gems log file
| saying
|
| -----------  Unreportable ERROR Encountered:
| 2012-09-22T11:19:06.81087493896484+02:00a ImproperOperation occurred
| (error 2089), reason:rtErrBadCopyFromTo, An index range was
| specified for a sequenceable collection with the starting index < a
| lot of ^@ characters > ( ...) greater than the ending index
| 0.-----------
|
| To this mail I attach a set of log files that contain everything from
| the start of the stone to the crash and side effects. Hope this
| helps. If not you need to guide me to test other things.
|
| Norbert
|
|
|
|  
| Am 20.09.2012 um 21:39 schrieb Dale Henrichs <[hidden email]>:
|
| >
| >
| > ----- Original Message -----
| > | From: "Norbert Hartl" <[hidden email]>
| > | To: "GemStone Seaside beta discussion"
| > | <[hidden email]>
| > | Sent: Thursday, September 20, 2012 9:04:23 AM
| > | Subject: Re: [GS/SS Beta] GemStone 3.1 problem opening debugger
| > |
| > | Dale,
| > |
| > |
| > | Am 20.09.2012 um 17:12 schrieb Dale Henrichs
| > | <[hidden email]>:
| > |
| > | > Norbert,
| > | >
| > | > If you want to look at the 3.1 extent from 3.1.0.1, then you
| > | > need
| > | > to upgrade the extent first.
| > | >
| > | > The upgrade instructions are in the installation guide.
| > | >
| > | Where can I find the 3.1.0.1 guide? I only have the 3.1.0 guide.
| >
| > The latest docs can always be found here[1]. Older versions of the
| > docs can be found on the sidebar...
| >
| > Dale
| >
| > [1]
| > http://community.gemstone.com/display/GSS64/GemStoneS+64+Documentation
| >
| >
|
|

Dale,

Am 24.09.2012 um 19:03 schrieb Dale Henrichs <[hidden email]>:

> Norbert,
>
> When you "open the debugger" are you opening the debugger against a continuation that was created in 2.4.4.1 or are you "opening the debugger" on a new continuation ... or ?

it seems I opened 3.1.0 continuations (I don't believe they are older) in the debugger when he went boom. Clearing the object log and producing new continuations after the upgrade lets me open a debugger again. I couldn't imagine it is this brittle when upgrading. So it seems after all I didn't do _every_ step that is necessary to do a clean upgrade. If I find further things I come back to you.

thanks,

Norbert