Warning: sendmail stops sending when load is high

[gormank]

This isn't a problem that needs support. Just FYI...

It seems sendmail has features that queue or disable sending mail if load gets high on the server. It seems it starts queuing mail when load (top, uptime) gets to 8 and refuses to send at 12. The defaults seem to match the commented lines in the .cf. Here's a message from maillog:

rejecting connections on daemon MTA: load average: 15

# grep eLA /etc/mail/sendmail.cf
#O QueueLA=8
#O RefuseLA=12

We had a bunch of servers down last night and for some reason that seems to have made Nagios repeatedly check to the point load got out of hand. Any suggestions for limiting repeated checks so this doesn't happen? I reduced the number of retries so that may take care of it.

[rkennedy]

The retries should take care of it, as well as modifying the two variables you mentioned in the sendmail.cf.

Code: Select all

# load average at which we just queue messages
#O QueueLA=8

# load average at which we refuse connections
#O RefuseLA=12

[gormank]

There won't be any retries from Nagios and until the load drops, sendmail just keeps refusing to send mail.
This was more an info post in case others see the issue, or developers decide to reconfigure sendmail as part of the installation.

[tmcdonald]

When you say Nagios would repeatedly retry when sendmail queued/refused to send, hoq quickly was it re-checking? Was it respecting the retry_interval setting? I can't think of anything that would cause it to act otherwise.

Regarding setting the defaults, we could possibly do that but I think it would be better to treat this in the same way as NPCD in that we recommend setting it to some fixed multiple of the number of cores in the system. Thoughts?

[gormank]

Nagios sent the notification to sendmail and is done with it, so Nagios isn't going to retry. The notifications are visible in Nagios under home, notifications.
The first time I ran into this issue was when I checked the system after not getting notified. Probably not the best way to find out. I know of no Nagios docs that say to adjust sendmail settings.

[rkennedy]

Thanks for letting us know. In your environment, have you modified the two variables above according to the amount of cores you have?