Hylafax Mailing List Archives
|
[Date Prev][Date Next][Thread Prev][Thread Next]
[Date Index]
[Thread Index]
Re: Faxmail error? HELP!!!
On 30-Sep-97 Doug White wrote:
>On Tue, 30 Sep 1997, Patrick Gardella wrote:
>
>> I am running Hylafax 4.0pl1 on FreeBSD 2.2.1 on a Pentium 200 with 128 Megs
>> RAM. Sendmail is 8.8.5. Modems are seven Multitech Multimodems 2834ZDX on a
Cy
>> clades8Yo multiport serial card. I'm sending this to both the HylaFAX and
>> FreeBSD lists to get input from both parties.
>
>OK, scratch login.conf.
>
>> All of them had the error code 2 that says "Cannot fork"
>>
>> I've included some stats on the system to provide some clues to what
>> might be happening.
>
>You're running out of processes per child.
>
>Try rebuilding your kernel with the following options:
>
>options CHILD_MAX=128
>options OPEN_MAX=128
>
>You might try just rebooting too. I'd have to see a ps -ax to get a sense
>of what's going on. hylafax may be leaking memory or file descriptors.
I've done this and will see what happens (kernel and reboot). It may take a
while to have this problem come up again, since this did not happen until
after almost 1 1/2 months of continuous running without a reboot.
>> The kernel is compiled with "maxusers" as 40 giving me 656 processes.
>
>That is somewhat loose; the above variables control this much more
>directly.
>
>Hm, there was one zombie process. You might want to keep track of those
>and make sure they don't get too numerous. A zombie process is one that's
>waiting for it's parent to recognize that it's dead. If big apps (like
>faxmail or anything fax related) is getting stuck, it doesn't release it's
>resources until it's postiviely dead. They're generally caused by
>programing errors related to not understanding the BSD process model.
The zombie processes have diappeared, by the time I went in to copy down
the ps -ax data. I did see that they were (hfaxd) and (faxq), before when
I first did ps -ax.
I'll keep everyone up to date.
Thanks Doug.
Patrick
P.S. Doug, is maxusers ok at 40? Or is there a better value?