Repo watch notification email issue and question -- Emails stop sending

Brian Richardson November 30, 2011

Hi,

We have been struggling with the Watch email notification feature for some time now. Here are my issues and concerns that I would greatly appreciate help from the community regarding.

I am using (Version:2.6.2 Build:20110720020604 2011-07-20), monitoring 16 branches that are a total of 1.5GB in size of code. The production server that hosts FishEye is a 48CPU 256 GB RAM Red Hat machine that only hosts FE and one other application as of now.

1) Email notifications stop sending out to users at arbitrary times. FishEye will run just fine but there is no way determine if email is down. I do see this in the log file at times and never recovers:

  • "[SvnTimeout ] fisheye.app SvnThrottledClient-doInterrupt - Interrupting SVN operation thread due to non-response: IncrementalPinger1:76"

SVN is always up and running so it is not an SVN inaccessiblity issue.

I then have to determine that I have not received an email in some time and bounce the server.

This results in users receving a flood of emails resulting in several managers pinging my team angrily

2) After bouncing the FishEye server the email queue recovery takes 8-10 hours

  • The slurping of the files happens quite efficiently and indexing eventually catches up
  • Our repos are moderatly busy but nothing out of the ordinary for checkin activity.
  • Yesterday FE email was down from 3:49am until 9pm and it took 9 hours to send the block of emails to the users.

1 answer

0 votes
seb
Atlassian Team
Atlassian Team members are employees working across the company in a wide variety of roles.
December 1, 2011

Hi Brian,

Most likely, your subversion indexing thread is being locked up when scanning large changesets and this is causing the emails to also get blocked. The "non-response" error in your logs is in our FishEye knowledgebase and is a known bug at FE-1071.

The first port of call would be follow the instructions on the knowledgebase article and possibly consider setting up svnsync to index from a local filesystem. If this does not help, then we always suggest upgrading FishEye to our latest release (currently 2.7.8) - although this particular issue has not been addressed, we are constantly improving and fixing issues in our indexing code, which may ameliorate this issue.

We understand that there may be more sinister factors at play and I would recommend that you raise a support request at https://support.atlassian.com if you are still having troubles. Our support engineers will help diagnose and resolve this problem. More detailed instructions here: http://confluence.atlassian.com/display/FISHEYE/Contacting+Support.

Regards,

Seb

Suggest an answer

Log in or Sign up to answer
TAGS
AUG Leaders

Atlassian Community Events