elliefm
commented
7 years ago If you have the opportunity (since the month is about to roll over again), are you able to reproduce this with the following patch in place?
diff --git a/imap/imapd.c b/imap/imapd.c
index 3dfb5bb..d73bc1c 100644
--- a/imap/imapd.c
+++ b/imap/imapd.c
@@ -12313,8 +12313,11 @@ static int getdatetime(time_t *date)
buf[i] = '\0';
r = time_from_rfc3501(buf, date);
- if (r < 0)
+ if (r < 0) {
+ syslog(LOG_DEBUG, "%s: time_from_rfc3501 rejected date string: '%s'",
+ __func__, buf);
goto baddate;
+ }
c = prot_getc(imapd_in);
return c;My hunch at the moment is that this is going to turn out to be some sort of bug in Cassandane, but it'd be very useful to see what's actually ending up in Cyrus
Yesterday evening, which for me was about 8PM CDT on May 31st, my normal cassandane run picked up 152 errors. All that I checked were of the form:
1) test_delete(Cassandane::Cyrus::ACL) Perl exception: IMAP Command : 'append' failed. Response was : bad - Invalid date-time in Append command
This morning, with no changes to my packaging, those have all gone away.
Now it's possible that one of the underlying dependencies changed in a buggy way and then was fixed overnight, but I have diffed the installed dependency list between the good and the bad builds and don't see any differences. It's also possible that Cassandane (or even the core Cyrus code, I guess) has some problems with dates near the end of the month. Maybe a timezone problem, where UTC is at the next month but local time isn't. I tried building with TZ=UTC explicitly but it made no difference.
I'm out of ideas so I figured I'd file a ticket just in case someone might be able to make a guess as to what might have happened.
Here's the output from just one of those failing tests, though there doesn't appear to be anything particularly useful in there.
And just in case it helps (though I don't think it will), this is what gets logged to syslog for that 023333A1 test: