Project

General

Profile

Bug #3269

idmapd hangs?

Added by Anton Anikin over 7 years ago. Updated over 4 years ago.

Status:
Closed
Priority:
Normal
Assignee:
-
Category:
-
Start date:
2012-10-09
Due date:
% Done:

0%

Estimated time:
Difficulty:
Medium
Tags:
needs-triage
Gerrit CR:

Description

Hello!
I have one server on OI_151a5, it's cifs server on local network joined to AD (windows 2003).
Sometimes users can't access their data on server (server asks for auth).
But after restart of idmap service everything works fine again.
So i think it idampd problem.
After idmap restart i see this in /var/adm/messages:

Oct 9 20:16:33 STORE smbd[663]: [ID 801593 daemon.error] idmap_get_mappings: RPC error
Oct 9 20:16:33 STORE smbd[663]: [ID 801593 daemon.error] smb_idmap_batch_getmappings: RPC error
Oct 9 20:16:33 STORE smbd[663]: [ID 266262 daemon.error] NT Authority\Anonymous: idmap failed
Oct 9 20:16:34 STORE smbd[663]: [ID 807464 daemon.error] ndr_rpc_bind: smbrdr_ctx_new(S=dc1, D=VFX, U=STORE$), err=131
Oct 9 20:16:34 STORE smbd[663]: [ID 807464 daemon.error] ndr_rpc_bind: smbrdr_ctx_new(S=dc1, D=VFX, U=STORE$), err=17
Oct 9 20:16:34 STORE last message repeated 33 times

I thought it was probably a memory leak.
When it happened again, I took a dump with gcore.
And ::findleaks in mdb show me this:
@
BYTES LEAKED VMEM_SEG CALLER
4096 33 f9de0000 MMAP
4096 1 fec68000 MMAP
16384 1 fec5b000 MMAP
4096 1 fdcee000 MMAP
4096 1 fd8fe000 MMAP
4096 1 fd22e000 MMAP
4096 1 fd00e000 MMAP
4096 1 fce0e000 MMAP
4096 1 fcd0f000 MMAP
4096 1 fc9fe000 MMAP
4096 1 fc8ff000 MMAP
4096 1 fc800000 MMAP
4096 1 fc701000 MMAP
4096 1 fc602000 MMAP
4096 1 fc503000 MMAP
4096 1 fc404000 MMAP
4096 1 fc305000 MMAP
4096 1 fc206000 MMAP
4096 1 fc107000 MMAP
4096 1 fc008000 MMAP
4096 1 fbf09000 MMAP
4096 1 fbcfe000 MMAP
4096 1 fbbff000 MMAP
4096 1 fbb00000 MMAP
4096 1 fba01000 MMAP
4096 1 fb902000 MMAP
4096 1 fb803000 MMAP
4096 1 fb704000 MMAP
4096 1 fa82e000 MMAP
4096 1 fa2fe000 MMAP
4096 1 fa0fe000 MMAP
4096 1 f9fde000 MMAP
4096 1 f9edf000 MMAP
82176 1 80f2560 0
------------------------------------------------------------------------
Total 34 oversized leaks, 229632 bytes

CACHE LEAKED BUFFER CALLER
080e1590 593 08128000 ?
080e2010 1199 08114d10 ?
080e2590 601 08134120 ?
080e3010 606 0811e0a8 ?
------------------------------------------------------------------------
Total 2999 buffers, 81736 bytes
@

In /var/svc/log/system-idmap:defaul.log:

Using server dc1.vfx.local:3268
Using server dc1.vfx.local:3268
thread creation refused - 40 threads currently active
thread creation refused - 40 threads currently active
[ Oct 9 20:16:33 Stopping because service restarting. ]
[ Oct 9 20:16:33 Executing stop method (:kill). ]
Terminating.
[ Oct 9 20:16:33 Executing start method ("/usr/lib/idmapd"). ]
created thread ID 3 - 1 threads currently active
[ Oct 9 20:16:33 Method "start" exited with status 0. ]
created thread ID 5 - 2 threads currently active
created thread ID 6 - 3 threads currently active

So I need help to debug this problem.
Thanks.


Related issues

Related to illumos gate - Bug #3282: idmap hangsClosed2012-10-17

Actions

History

#1

Updated by Gordon Ross over 7 years ago

  • Status changed from New to Feedback

This is not a "site" issue - should close.

#2

Updated by Rich Lowe about 7 years ago

  • Project changed from site to illumos gate
#3

Updated by Gordon Ross over 4 years ago

  • Status changed from Feedback to Closed

The problem described is what it looks like when your AD server becomes unresponsive.
After #6352 we recover much better from a DC going away, so let's say that fixed this.

Also available in: Atom PDF