Long Lived imapd processes
Earl R Shannon
Earl_Shannon at ncsu.edu
Tue Apr 15 10:47:14 EDT 2003
The processes are using resources on a resource limited machine. We are
looking at all the ways we can reduce the load on the machine. Having a
client connected for a month and not reading mail is not productive for
Ken Murchison wrote:
> Earl R Shannon wrote:
>>Running Cryus IMAP V 2.1.11 on Solaris 7
>>We have been finding long lived imapd processes on our imap server. Some
>>approaching thirty days. A truss of these processes show that they are
>>not doing anything except receiving a noop from the client every ten
>>minutes or so. This resets the timer which allows the process to remain.
> Right. Somebody has a client just sitting there polling the server.
> Why is this a problem? As long as the server responds, its doing its
>>My question is: what if anything do others on this list do about these
>>situations? I am concerned about killing imapd processes and what it may
>>eventually do to the master process.
> As it currently stands, if you kill an imapd process otu from under
> master, master will think it is still active and will assume that it can
> service requests. If you do this enough, master won't spawn any more
> imapd and no clients will be serviced.
More information about the Info-cyrus