|
Problem IT33289
|
Status: Closed |
IF INSTANCE HAS SMART BLOB SPACES, STARTING 2 ARCHIVES CAN HANG SERVER IN BLOCKED:ARCHIVE WAITING FOR E0SB_ARCSPACE MUTEX
|
| product: |
INFORMIX SERVER / 5725A3900 / C10 - IDS 12.10 |
| Problem description: |
This problem was reported against an IDS server running a
special build of 12.10.FC7W1 on linux x86-64
and I confirmed the problem exists in 12.10.FC14 as well.
If your IDS instance has a smart blob space and you start 2
concurrent archives, it's possible for
the server to enter a continual hang blocking all sessions from
entering critical sections.
When the problem occurs, your onstat banner will display
"Blocked:ARCHIVE"
You will see two ontape threads in "onstat -g ath" output like
this:
$ onstat -g ath | grep ontape
30026 47c250e8 44da5a10 2 cond wait
block 8cpu ontape
30231 479ca328 44da2c98 2 mutex wait
E0sb_arcsp 1cpu ontape
An "onstat -g lmx" will show the following locked mutexes:
$ onstat -g lmx | head
IBM Informix Dynamic Server Version 12.10.FC14 -- On-Line --
Up 02:13:01 -- 218964 Kbytes
Blocked:ARCHIVE
Locked mutexes:
mid addr name holder lkcnt
waiter waittime
2953 44bdfce8 B3sb_arcspace 30231 0
2990 44be5928 E0sb_arcspace 30026 0
30231 1461
And stack traces for your ontape threads will look like:
Stack for thread: 30026 ontape
base: 0x0000000047247000
len: 69632
pc: 0x0000000001bf6968
tos: 0x0000000047252050
state: cond wait
vp: 8
0x0000000001bf6968 (oninit) yield_processor_mvp
0x0000000001bf3253 (oninit) mt_fast_wait
0x0000000000f7e13d (oninit) critblockflags
0x0000000000f7e6e2 (oninit) critwait
0x0000000000ff5fa7 (oninit) critenter
0x0000000001087670 (oninit) stmp_drop
0x0000000001a9ce27 (oninit) sbx_arc_end
0x00000000011d0919 (oninit) arc_sblob_cleanup
0x00000000011c6ad0 (oninit) init_arcbu
0x00000000011c1eca (oninit) init_archive
0x00000000011be002 (oninit) isopen_arcbu
0x000000000087d53d (oninit) sqisopen_arcbu
0x0000000000dd1d22 (oninit) tbj_open_archive
0x0000000000dd186e (oninit) tbj_archive
0x0000000000d33c31 (oninit) sq_archive
0x0000000000d2b8d7 (oninit) sqmain
0x0000000001da43d1 (oninit) listen_verify
0x0000000001da340c (oninit) spawn_thread
0x0000000001b8b297 (oninit) th_init_initgls
0x0000000001c15849 (oninit) startup
Stack for thread: 30231 ontape
base: 0x0000000047bed000
len: 69632
pc: 0x0000000001bf6968
tos: 0x0000000047bf81b0
state: mutex wait
vp: 1
0x0000000001bf6968 (oninit) yield_processor_mvp
0x0000000001bcd110 (oninit) mt_lock_wait
0x0000000001bce2d0 (oninit) mt_lock_helper
0x0000000001bcdee5 (oninit) mt_lock
0x0000000001b83d5b (oninit) gtm_hash_get_wlock
0x0000000001a9bd07 (oninit) sba_arcspace_get_wlock
0x0000000001a9cb89 (oninit) sbx_arc_uelists_bld
0x00000000011d04f1 (oninit) arc_bld_sblob_uelists
0x00000000011c341c (oninit) init_archive
0x00000000011be002 (oninit) isopen_arcbu
0x000000000087d53d (oninit) sqisopen_arcbu
0x0000000000dd1d22 (oninit) tbj_open_archive
0x0000000000dd186e (oninit) tbj_archive
0x0000000000d33c31 (oninit) sq_archive
0x0000000000d2b8d7 (oninit) sqmain
0x0000000001da43d1 (oninit) listen_verify
0x0000000001da340c (oninit) spawn_thread
0x0000000001b8b297 (oninit) th_init_initgls
0x0000000001c15849 (oninit) startup
Description |
| Problem Summary: |
****************************************************************
* USERS AFFECTED: *
* Users of Informix 12.10.14 and earlier releases. *
****************************************************************
* PROBLEM DESCRIPTION: *
* If your IDS instance has a smart blob space and you start 2 *
* concurrent archives, it's possible for *
* the server to enter a continual hang blocking all sessions *
* from *
* entering critical sections. *
****************************************************************
* RECOMMENDATION: *
**************************************************************** |
| Local Fix: |
|
| Solution |
|
| Workaround |
****************************************************************
* USERS AFFECTED: *
* Users of Informix 12.10.14 and earlier releases. *
****************************************************************
* PROBLEM DESCRIPTION: *
* If your IDS instance has a smart blob space and you start 2 *
* concurrent archives, it's possible for *
* the server to enter a continual hang blocking all sessions *
* from *
* entering critical sections. *
****************************************************************
* RECOMMENDATION: *
**************************************************************** |
| Comment |
Upgrade to Informix 12.10.xC15 or later. |
| Timestamps |
Date - problem reported :
Date - problem closed :
Date - last modified :
| 22.06.2020
29.12.2020
29.12.2020 |
| Problem solved at the following versions (IBM BugInfos) |
|
| Problem solved according to the fixlist(s) of the following version(s) |
