Logging missing large chunk of the day?

[Scottyjohn]

Hi all,
We are running Framework 7.2 and have just started getting a problem which is very odd.  On our pool of statservers the logging is stopping at roughly the same time around 1900hrs every night and not restarting logging till about 1300hrs the next day?  All start stop times are not sunchronised, i.e. they happen randomly but roughly at the same point in the day.
All other apps seem to be logging normally.

Any ideas what could be happening?

[Fra]

Mmh..are they running on the same box?

[Scottyjohn]

No, they are split primary and back up laternating across 2 servers

[Scottyjohn]

Sorry should read across alternating servers 

[victor]

Well, there are quite a few things I can imagine going wrong here (other than lactating servers)

First of all, how many old files do you keep before deleting them? Is it my number of files or by date?
Also, how often do your statservers spew out new files?

And finally, did you check release notes for any problems in 7.2 statserver regarding logging? Sounds like the easiest thing to start with.
Vic

[mike kamlet]

Do you have any "log scripts" that move files around? compress? etc??

Are folks actively reading the log files at the time of the missing data?

[Scottyjohn]

Hmm, I have checked the release notes for our version 7.6.100.12 and there is nothing there about issues with logging.  Ive attached a screenshot which shows the logging for one of our statservers and you can see that there is a large jump at the same point each day.  There is a script which compresses the logs as you can see from the screenshot

[img width=488 height=480]http://farm4.static.flickr.com/3577/3287675232_8709392204_o.jpg[/img]


Any thoughts would be appreciated

[cavagnaro]

Can you disable that script to test?

[KBS]

When does that script kick off? Why are you compressing the files?
I'm not sure if this applies for Stat Server, but we've had an issue lately with URS "stalling" for a 12 - 30 seconds in between certain (not every) log files.
When I asked Genesys about it, they suggested I :
- make sure that the URS was the only application writing to that directory, since the application scans the directory each time a new log file is created;
- I was not using any compression on the folder, because this adds to disk I/O and processor time
- examine Disk I/O

Granted, this is a Windows system and not Unix, but...

Question:
What shows in the log files? Is there continuity within those files - for example, is the end of one file just before the start of the next? Do you have the checkpoint option enabled in your logging?  Does the traffic on that Stat Server decrease around the end of the day? For example, if that is feeding CC Pulse only, are there less users making statistic requests at the end of the day, compared to earlier? What Stat Server logging options do you have enabled? Are there any other processes (host backup, etc) that could be hoarding processor time?

[manas]

Looks like a case of stat server stalling and going into a hung state.  The interesting part is it going to hung state at same time which may be just coincidence or there can be other factors. It was happening in our environment too  and Genesys recommended  upgrading  to Stat Server 7.6.100.19 AIX 32/64.