Smalltalk › Gemtalk › GLASS

[Glass] tranlog generation speed

_‹ Previous Topic Next Topic _›

Classic

List

Threaded

2 messages Options

leodm

[Glass] tranlog generation speed

hi all!

I'm finally have a really nice cleanup script and backup running in crontab for maintenance.

The scripts run ok for weeks but now I appreciated that the tranlog generation is too fast for the cleanup script,

for example my script delete all tranlog files older than 3 days but I have a 16GB disk for tranlog files only and only

in half day already in 50% used!

Obviusly I can change the script to be more accurated but is normal this bitrate generation for tranlogs?

thnks in advance

Leo

_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

Dale Henrichs-3

[Glass] Fwd: tranlog generation speed

new mailer and replied to sender instead of list ...

---------- Forwarded message ----------
From: Dale Henrichs <[hidden email]>
Date: Wed, Jan 15, 2014 at 12:07 PM
Subject: Re: [Glass] tranlog generation speed
To: [hidden email]

Leo,

First I just want to clarify. since you don't mention your backup schedule, I just want to be complete:)

You say that you are deleting tranlogs every 3 days and I am curious how often you do a backup ... At a minimum you need to keep all of the tranlogs created since a backup plus a fudge factor of tranlogs prior to backup (there's a call to find out exactly which tranlogs are needed ... I can supply this if you aren't already doing so) ...

The tranlogs record information about what is going in your system, so there is no "global normal" for tranlog size ... If you see a change in the number/size of tranlogs created during a standard period of time, then I would expect that something "abnormal" has gone on in your system where you are performing more transactions and/or recording larger amounts of data as persistent ...

The "abnormality" could be increased traffic and related to that or the "abnormality" could be related to the object log and continuations ... continuations can reference large hunks of "normally non-persistent objects" and that would lead to abnormally large tranlogs ...

Along with larger tranlog sizes, I would also expect to see a growth in your extent size ... depending upon whether or not you ar running epoch gc and the grequency of your MFC runs ...

Dale

On Wed, Jan 15, 2014 at 11:51 AM, <[hidden email]> wrote:

hi all!

I'm finally have a really nice cleanup script and backup running in crontab for maintenance.

The scripts run ok for weeks but now I appreciated that the tranlog generation is too fast for the cleanup script,

for example my script delete all tranlog files older than 3 days but I have a 16GB disk for tranlog files only and only
in half day already in 50% used!

Obviusly I can change the script to be more accurated but is normal this bitrate generation for tranlogs?

thnks in advance
Leo

_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass

_______________________________________________
Glass mailing list
[hidden email]
http://lists.gemtalksystems.com/mailman/listinfo/glass