Dec 23 11:05:44 gehenna18.rutgers.edu dovecot: IMAP(user1): Panic: file
maildir-uidlist.c: line 403 (maildir_uidlist_records_array_delete):
assertion failed: (pos != NULL)
This is hopefully fixed by:
http://hg.dovecot.org/dovecot-1.2/rev/d00abdaf9df2
Thanks for the help as always. I'm currently patching 1.2.9 with this
and pushing the changes out to our dovecot servers. I'll update you
later today if we stop seeing the panic.
Dec 23 11:18:32 gehenna17.rutgers.edu dovecot: IMAP(user2): Panic: file
cmd-list.c: line 242 (ns_get_listed_prefix): assertion failed: (match ==
IMAP_MATCH_YES)
So you get a core dump? Could you print a few values:
fr 6 (or whatever starts giving the right values)
p *ctx
p *ctx.ns
p ctx.patterns[0]
p ctx.patterns[1]
p ctx.patterns[2]
p match
p ns_prefix
p p
The ctx.patterns values are the most important ones I think.
The values from lines 5-7 all look the the same. Here's the answers from
line 6 though:
(gdb) fr 6
#6 0x000000000041c9c2 in cmd_list_continue (cmd=0x1212dcc8) at
cmd-list.c:375
375 list_namespace_send_prefix(ctx, TRUE);
(gdb) p *ctx
$17 = {cmd = 0x1212dcc8, ref = 0x12132b50 "", patterns = 0x1212de48,
list_flags = 16386, status_items = 0, ns = 0x1212bf70, list_iter =
0x121304f8, ns_prefixes_listed = {arr = {buffer = 0x1212de58,
element_size = 8}, v = 0x1212de58,
v_modifiable = 0x1212de58}, lsub = 0, lsub_no_unsubscribed = 0,
inbox_found = 1, seen_inbox_namespace = 1, cur_ns_match_inbox = 1,
cur_ns_send_prefix = 0, cur_ns_skip_trailing_sep = 1, used_listext = 0}
(gdb) p *ctx.ns
$18 = {next = 0x0, type = NAMESPACE_PRIVATE, sep = 46 '.', real_sep = 46
'.', sep_str = ".\000", flags = 8213, prefix = 0x1212bfd0 "INBOX.",
prefix_len = 6, alias_for = 0x0, alias_chain_next = 0x0, user = 0x1212b988,
owner = 0x1212b988, list = 0x1212c348, 0x1212c038}
(gdb) p ctx.patterns[0]
$19 = 0x12132b58 "Inbox"
(gdb) p ctx.patterns[1]
$20 = 0x0
(gdb) p ctx.patterns[2]
$21 = 0x1212de90 ""
(gdb) p match
No symbol "match" in current context.
(gdb) p ns_prefix
No symbol "ns_prefix" in current context.
(gdb) p p
No symbol "p" in current context.
...and here's the full trace for reference
http://pastebin.com/f77189785
--
================================
David Halik
System Administrator
OIT-CSS Rutgers University
dha...@jla.rutgers.edu
================================
