[tode_st] tODE in client machine or X11 forward?

Previous Topic Next Topic
 
classic Classic list List threaded Threaded
8 messages Options
Reply | Threaded
Open this post in threaded view
|

[tode_st] tODE in client machine or X11 forward?

Mariano Martinez Peck
Hi guys,

For GemTools there was no discussion that gemtools installed on server and X11 forward was MUUUUUCH faster than gemtools in client machine pointing a remote stone.  

Now...for tODE does someone had tested both alternatives? Is X11 still the fastest/recommended way? 

Cheers, 

--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: [tode_st] tODE in client machine or X11 forward?

Dale Henrichs-3
From a docs perspective we will be covering this in a bit more detail. Here's an (out-of-date) starting point[1] that references the x11 and netldi port forwarding options ...

I will be splitting the client structure away from the server structure in the next couple of weeks ... this will make it easier to start using Pharo4.0 or Pharo5.0 without having to do major conniption fits ... also it will make the windows installation much easier .... so the docs will lag on this stuff until I get the projects split up...

At home I use a VPN to connect directly to my netldi, and it performs very well.

I tend to recommand using the netldi port forwarding,even though the x11 forwarding is better with the compression I still see certain operations perform slowly ... So i'm biased towards having the client run natively...

I guess you could try both ways and see which one you like ... if you have the client installed remote from your server, then the descirption files need to be synched between both environments, but that's not too hard to do ..

there are also things like X2Go(?) that are apparently pretty performant, but they require that you have a desktop installation on the server as x2go exports the whole desktop (I think) ...

Dale

[1] https://github.com/GsDevKit/gsDevKitHome/blob/dev/docs/clientInstallation.md#tode-client-installation

On 09/14/2015 02:57 PM, Mariano Martinez Peck wrote:
Hi guys,

For GemTools there was no discussion that gemtools installed on server and X11 forward was MUUUUUCH faster than gemtools in client machine pointing a remote stone.  

Now...for tODE does someone had tested both alternatives? Is X11 still the fastest/recommended way? 

Cheers, 
--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: [tode_st] tODE in client machine or X11 forward?

Mariano Martinez Peck


On Mon, Sep 14, 2015 at 7:18 PM, Dale Henrichs <[hidden email]> wrote:
From a docs perspective we will be covering this in a bit more detail. Here's an (out-of-date) starting point[1] that references the x11 and netldi port forwarding options ...

I will be splitting the client structure away from the server structure in the next couple of weeks ... this will make it easier to start using Pharo4.0 or Pharo5.0 without having to do major conniption fits ... also it will make the windows installation much easier .... so the docs will lag on this stuff until I get the projects split up...

At home I use a VPN to connect directly to my netldi, and it performs very well.


OK, I think I will start with netldi port forwarding. But I will need to first change all netldi ports for the existing stones...
 
I tend to recommand using the netldi port forwarding,even though the x11 forwarding is better with the compression I still see certain operations perform slowly ... So i'm biased towards having the client run natively...


OK.. just for the record, did you try the compression thingy for the port forwarding for your VPN. Something like:

ssh -c arcfour,blowfish-cbc -XC  -N -L 40101:localhost:40101 -L 40102:localhost:40102 -L 40103:localhost:40103   yourserver

I just tried and didn't see much difference...but just wondering...
 
I guess you could try both ways and see which one you like ... if you have the client installed remote from your server, then the descirption files need to be synched between both environments, but that's not too hard to do ..

there are also things like X2Go(?) that are apparently pretty performant, but they require that you have a desktop installation on the server as x2go exports the whole desktop (I think) ...

Ok, thanks. I will start with netldi port forwarding. Thanks!

 

Dale

[1] https://github.com/GsDevKit/gsDevKitHome/blob/dev/docs/clientInstallation.md#tode-client-installation


On 09/14/2015 02:57 PM, Mariano Martinez Peck wrote:
Hi guys,

For GemTools there was no discussion that gemtools installed on server and X11 forward was MUUUUUCH faster than gemtools in client machine pointing a remote stone.  

Now...for tODE does someone had tested both alternatives? Is X11 still the fastest/recommended way? 

Cheers, 
--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.

--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.



--

--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: [tode_st] tODE in client machine or X11 forward?

Dale Henrichs-3


On 09/15/2015 12:02 PM, Mariano Martinez Peck wrote:


On Mon, Sep 14, 2015 at 7:18 PM, Dale Henrichs <[hidden email]> wrote:
From a docs perspective we will be covering this in a bit more detail. Here's an (out-of-date) starting point[1] that references the x11 and netldi port forwarding options ...

I will be splitting the client structure away from the server structure in the next couple of weeks ... this will make it easier to start using Pharo4.0 or Pharo5.0 without having to do major conniption fits ... also it will make the windows installation much easier .... so the docs will lag on this stuff until I get the projects split up...

At home I use a VPN to connect directly to my netldi, and it performs very well.


OK, I think I will start with netldi port forwarding. But I will need to first change all netldi ports for the existing stones...
 
I tend to recommand using the netldi port forwarding,even though the x11 forwarding is better with the compression I still see certain operations perform slowly ... So i'm biased towards having the client run natively...


OK.. just for the record, did you try the compression thingy for the port forwarding for your VPN. Something like:

ssh -c arcfour,blowfish-cbc -XC  -N -L 40101:localhost:40101 -L 40102:localhost:40102 -L 40103:localhost:40103   yourserver

I just tried and didn't see much difference...but just wondering...
I did not do that ... after writing the email, I thought that I should try it ...:)
 
I guess you could try both ways and see which one you like ... if you have the client installed remote from your server, then the descirption files need to be synched between both environments, but that's not too hard to do ..

there are also things like X2Go(?) that are apparently pretty performant, but they require that you have a desktop installation on the server as x2go exports the whole desktop (I think) ...

Ok, thanks. I will start with netldi port forwarding. Thanks!

Yeah when doing the port forwarding, sometimes the Transcript shows: (like during a large load) can slow things down, because each show is a roundtrip (or two). There is a class variable in TDTopezGemStoneClient where you can turn off the transcript show traffic:

    TDTopezGemStoneClient slowConnection: true
    TDTopezGemStoneClient slowConnection: false

when `true` the transcript show traffic is turned off ... but you have to set the env var BEFORE login for it to take effect ... the transcript show's are still recording in object log/gem log, they just aren't routed over the wire ...

Like I said, I prefer the netldi port forwarding to X11, but your results could be different:)

Dale
Dale

--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: [tode_st] tODE in client machine or X11 forward?

Mariano Martinez Peck


On Tue, Sep 15, 2015 at 4:23 PM, Dale Henrichs <[hidden email]> wrote:


On 09/15/2015 12:02 PM, Mariano Martinez Peck wrote:


On Mon, Sep 14, 2015 at 7:18 PM, Dale Henrichs <[hidden email]> wrote:
From a docs perspective we will be covering this in a bit more detail. Here's an (out-of-date) starting point[1] that references the x11 and netldi port forwarding options ...

I will be splitting the client structure away from the server structure in the next couple of weeks ... this will make it easier to start using Pharo4.0 or Pharo5.0 without having to do major conniption fits ... also it will make the windows installation much easier .... so the docs will lag on this stuff until I get the projects split up...

At home I use a VPN to connect directly to my netldi, and it performs very well.


OK, I think I will start with netldi port forwarding. But I will need to first change all netldi ports for the existing stones...
 
I tend to recommand using the netldi port forwarding,even though the x11 forwarding is better with the compression I still see certain operations perform slowly ... So i'm biased towards having the client run natively...


OK.. just for the record, did you try the compression thingy for the port forwarding for your VPN. Something like:

ssh -c arcfour,blowfish-cbc -XC  -N -L 40101:localhost:40101 -L 40102:localhost:40102 -L 40103:localhost:40103   yourserver

I just tried and didn't see much difference...but just wondering...
I did not do that ... after writing the email, I thought that I should try it ...:)
 
I guess you could try both ways and see which one you like ... if you have the client installed remote from your server, then the descirption files need to be synched between both environments, but that's not too hard to do ..

there are also things like X2Go(?) that are apparently pretty performant, but they require that you have a desktop installation on the server as x2go exports the whole desktop (I think) ...

Ok, thanks. I will start with netldi port forwarding. Thanks!

Yeah when doing the port forwarding, sometimes the Transcript shows: (like during a large load) can slow things down, because each show is a roundtrip (or two). There is a class variable in TDTopezGemStoneClient where you can turn off the transcript show traffic:

    TDTopezGemStoneClient slowConnection: true
    TDTopezGemStoneClient slowConnection: false

when `true` the transcript show traffic is turned off ... but you have to set the env var BEFORE login for it to take effect ... the transcript show's are still recording in object log/gem log, they just aren't routed over the wire ...


This is a very convenient method but still, it's not that much useful for my point of view if I have to log out and log in to take effect. In fact, I would like to disable transcript printing even for parts of code, or to block closures or only to one script. 
There is no way for this to take effect immediately without logging out and log in again? Can't we  do something similar as we did for the skipping of object log writing (remember we added a flag in the gem session cache?)? 

Or it would be cool to be able to do in tODE shell:

./updateQuuve > /dev/null 

:)


If not....do you remember how can I turn off those transcript show of Metacello?  

"Project: Gofer stable
Project Zinc Project baseline
...."

because it takes a lifetime to update my app code hahahhaa. 

Metacello >> #silently  does not do it...
  

Like I said, I prefer the netldi port forwarding to X11, but your results could be different:)

Dale
Dale

--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.



--

--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: [tode_st] tODE in client machine or X11 forward?

Dale Henrichs-3


On 09/17/2015 09:40 AM, Mariano Martinez Peck wrote:


On Tue, Sep 15, 2015 at 4:23 PM, Dale Henrichs <[hidden email]> wrote:


This is a very convenient method but still, it's not that much useful for my point of view if I have to log out and log in to take effect. In fact, I would like to disable transcript printing even for parts of code, or to block closures or only to one script. 
There is no way for this to take effect immediately without logging out and log in again? Can't we  do something similar as we did for the skipping of object log writing (remember we added a flag in the gem session cache?)? 

Or it would be cool to be able to do in tODE shell:

./updateQuuve > /dev/null 

:)


If not....do you remember how can I turn off those transcript show of Metacello?  

"Project: Gofer stable
Project Zinc Project baseline
...."

because it takes a lifetime to update my app code hahahhaa. 

Metacello >> #silently  does not do it...
 
Okay this is worth a feature request .... I could create a command that disables and enables the Transcript dynamically ... the SlowConnection: was something I did as a quick hack (which is why it isn't documented anywhere) ... we can do better:) I can imagine several flavors ... the dumping of logging info during a build is one of the most annoying transcripters so it would be useful to have a setting of some that in addition to the explict command would turn off the transcript during `project load ...` ... if there is a build error `ol view --age=`1 hours` could be used to look for failure clues ...

For your Transcript needs, if you turn off the client side Transcript with `TDTopezGemStoneClient slowConnection: false` ... is it too inconvenient to use the `ol view --age=`1 hours` -r transcript` to look at the transcript --- after the fact? Or do you use the transcript for watching progress in real time?


Dale



--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: [tode_st] tODE in client machine or X11 forward?

Mariano Martinez Peck


On Thu, Sep 17, 2015 at 3:13 PM, Dale Henrichs <[hidden email]> wrote:


On 09/17/2015 09:40 AM, Mariano Martinez Peck wrote:


On Tue, Sep 15, 2015 at 4:23 PM, Dale Henrichs <[hidden email]> wrote:


This is a very convenient method but still, it's not that much useful for my point of view if I have to log out and log in to take effect. In fact, I would like to disable transcript printing even for parts of code, or to block closures or only to one script. 
There is no way for this to take effect immediately without logging out and log in again? Can't we  do something similar as we did for the skipping of object log writing (remember we added a flag in the gem session cache?)? 

Or it would be cool to be able to do in tODE shell:

./updateQuuve > /dev/null 

:)


If not....do you remember how can I turn off those transcript show of Metacello?  

"Project: Gofer stable
Project Zinc Project baseline
...."

because it takes a lifetime to update my app code hahahhaa. 

Metacello >> #silently  does not do it...
 
Okay this is worth a feature request .... I could create a command that disables and enables the Transcript dynamically ... the SlowConnection: was something I did as a quick hack (which is why it isn't documented anywhere) ... we can do better:)

Yes, that would be nice. I let you open the issue since you can detailed better. 

Yes, there can be many flavors. I was thinking about 2:

1) In the same way you added disable and enable methods to object log logging from Transcript, we could also add similar ones for writing into the transcript. And since we have only 2 ways of logging right now (object log and transcript) just disabled both would be a null transcripter. 

2) We can very easily have a TranscriptProxy subclass called NullTranscriptProxy whose #show: does nothing and then having methods to enable / disable methods which would at least enable/disable transcript for the current gem..this could easily be done by doing "at:  #'TranscriptProxy put: NullTranscriptProxy "  in the transient symbollist of the session.  

 
I can imagine several flavors ... the dumping of logging info during a build is one of the most annoying transcripters so it would be useful to have a setting of some that in addition to the explict command would turn off the transcript during `project load ...` ... if there is a build error `ol view --age=`1 hours` could be used to look for failure clues ...

For your Transcript needs, if you turn off the client side Transcript with `TDTopezGemStoneClient slowConnection: false` ... is it too inconvenient to use the `ol view --age=`1 hours` -r transcript` to look at the transcript --- after the fact? Or do you use the transcript for watching progress in real time?

 
No, that is not too inconvinient. What is inconvinient is having to exit the shell and re enter to have effect (which is what I understood from your comment)

 

Dale



--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.



--

--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.
Reply | Threaded
Open this post in threaded view
|

Re: [tode_st] tODE in client machine or X11 forward?

Dale Henrichs-3


On 09/17/2015 11:21 AM, Mariano Martinez Peck wrote:


On Thu, Sep 17, 2015 at 3:13 PM, Dale Henrichs <[hidden email]> wrote:


On 09/17/2015 09:40 AM, Mariano Martinez Peck wrote:


On Tue, Sep 15, 2015 at 4:23 PM, Dale Henrichs <[hidden email]> wrote:


This is a very convenient method but still, it's not that much useful for my point of view if I have to log out and log in to take effect. In fact, I would like to disable transcript printing even for parts of code, or to block closures or only to one script. 
There is no way for this to take effect immediately without logging out and log in again? Can't we  do something similar as we did for the skipping of object log writing (remember we added a flag in the gem session cache?)? 

Or it would be cool to be able to do in tODE shell:

./updateQuuve > /dev/null 

:)


If not....do you remember how can I turn off those transcript show of Metacello?  

"Project: Gofer stable
Project Zinc Project baseline
...."

because it takes a lifetime to update my app code hahahhaa. 

Metacello >> #silently  does not do it...
 
Okay this is worth a feature request .... I could create a command that disables and enables the Transcript dynamically ... the SlowConnection: was something I did as a quick hack (which is why it isn't documented anywhere) ... we can do better:)

Yes, that would be nice. I let you open the issue since you can detailed better. 

https://github.com/dalehenrich/tode/issues/205

Dale

--
You received this message because you are subscribed to the Google Groups "tODE" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [hidden email].
For more options, visit https://groups.google.com/d/optout.