[EMAIL PROTECTED] wrote:
> A scan of the sources reveals that that error message is generated
> from the follow C code:
>
> if (chdir("queue") == -1) die(62);
>
> The reasons why that could fail are pretty limited in the qmail
> scenario.
>
> o The directory does not exist - installation error?
make setup
> o The file system is flaky - fsck?
I tested on three different machines (netra, E220R, E250)
- all seem to be quite stable apart from this issue.
> o The queue is on an NFS server - not allowed with qmail
Nightmare Filesystem - didn't use that for serious work
for the last year.
> o The process does not have permission - installation error?
make setup
> Hmmm, that's all I can think of right now.
>
> I have not heard of chdir intermittantly working before. Solaris or
> otherwise.
Hmm.
Today I acquired another E250 but under Solaris 2.6.
I installed an unpatched version of qmail - using an
umirrored disk for the queue -> Success - no error.
Reassured I installed the patched version with all the
nice features (conf-spawn=2045, conf-split=521) -> Success
- no error.
BTW.: I just edited /usr/include/sys/select.h (changing
1024 to 4096), I didn't edit /etc/system. Ulimit on that
machine says:
core file size (blocks) unlimited
data seg size (kbytes) 2097148
file size (blocks) unlimited
open files 64
pipe size (512 bytes) 10
stack size (kbytes) 8192
cpu time (seconds) unlimited
max user processes 16021
virtual memory (kbytes) unlimited
Another hint could be the fact, that the mails, which
remain in the queue after the first crash seem to be stuck
in there - they just don't get delivered. Mails of
following tests will be delivered completely - even if
qmail barfs. On the other hand - maybe they will be
delivered and I'll see that tomorrow ...
By
T�ns
--
Linux. The dot in /.
