notification.log is large

[Peter Heggie]

In 6.2, under ../integrator/server/logs, there is a file called notifications.log. It is over 500M in size. I copied /dev/null into it (this is a test server) and the length went to 0. In about a minute it jumped back up to the same size again.

What is the file used for and why is it suddenly growing? Have not seen this before. We have been running 6.2.0.2 on this server for two months; this is the first time I’ve seen this.

Peter Heggie

Topic

[Michael Hertel]

Thanks for pointing this out. I have never seen this before either.

I looked at my notification.log and it kept complaining about a site that was defined in the server.ini that was an old/unused/renamed site. So my log was growing but not at your rate.

I edited the server.ini to remove the offending site then recycled the hostserver which cycled the log. Hopefully all should be good now.

What is your log complaining about?  🙂

[Peter Heggie]

it was the same thing. The file was so big I could not open it for editing. I should have used the tail command to look at a sample. I edited the server.ini and it looks better now. In fact, it fixed the problem.

Thank you for looking at it!

Peter Heggie

[Peter Heggie]

Had issue again today. Looks like after an HA failover, when the HA scripts put all the production site names in the environs site list on the Test server, the HA scripts do not remove them when we failed back.

So meanwhile the host server was trying to connect to all these Prod sites that were no longer on the Test server. The notification.log file was up to 50M and the host server stopped. The host server log showed no errors but it was not running.

I removed the prod site names from environs in server.ini and bounced host server and no more error messages.

Shouldn’t a failback process include removing the prod sites from the test site list?

And shouldn’t the server.log have an error message before it stops?

Peter Heggie

[Author]

Replies