Red Hat Bugzilla – Bug 444321
yum update hangs -- FUTEX_WAIT
Last modified: 2014-01-21 18:02:33 EST
On a current rawhide x86_64 box, yum update hangs:
[root@scratchmonkey ~]# yum clean all
Loaded plugins: refresh-packagekit
Cleaning up Everything
[root@scratchmonkey ~]# yum update
Loaded plugins: refresh-packagekit
livna-development | 2.1 kB 00:00
primary.sqlite.bz2 | 118 kB 00:00
fedora | 2.4 kB 00:00
primary.sqlite.bz2 | 7.6 MB 00:01
pptp-stable | 1.9 kB 00:00
primary.sqlite.bz2 | 5.2 kB 00:00
adobe-linux-i386 | 951 B 00:00
primary.xml.gz | 10 kB 00:00
adobe-linu: ################################################## 17/17
updates | 2.4 kB 00:00
primary.sqlite.bz2 | 7.6 MB 00:02
Setting up Update Process
strace shows it sitting forever on
futex(0x7fc2fa534078, FUTEX_WAIT, 2, NULL
Changing version to '9' as part of upcoming Fedora 9 GA.
More information and reason for this action is here:
it's very old bug, I watch them on fc5, fc6, fc7, f8 and now on f9.
check bug 145021 which still present.
what information I need to catch to help to resolve this problem?
[timon@localhost ~]$ ps aux | grep yum
root 7303 0.0 4.0 61436 20828 ? S May29 0:01 /usr/bin/python
timon 8157 0.0 0.1 5068 760 pts/2 S+ 00:17 0:00 grep yum
[timon@localhost ~]$ sudo strace -p 7303
Process 7303 attached - interrupt to quit
futex(0xb7d9e0d0, FUTEX_WAIT, 2, NULL
There are two causes of FUTEX_WAIT that need to be separated if you wish a solution.
The 2 causes are
1) stale lock in Berkeley DB environment left registered when process exited
without unregistering the lock in the dbenv. A registered but abandoned lock
can be displayed/verified by running (see Berkely db_stat documentation):
cd /var/lib/rpm && /use/lib/rpm/rpmdb_stat -CA
and checking whether the processes/threads that own the lock(s) are still active.
FWIW, the check (and the cleanup) are largely automated for about 1.5 years now.
2) A stale FUTEX caused by process/thread termination without releasing the FUTEX.
This is a much rarer condition, and a more complicated diagnosis. There is now a
mechansim to handle reaping stale FUTEX's called"robust mutexes" that can clear
a stale FUTEX. The problem is that clearing a stale FUTEX is not sufficient, one
also has to deal with whatever state the lock was protecting in order to clear.
Luckily (for rpm) there is little important state that needs to be cleared in a rpmdb,
so reaping a stale FUTEX prevents the futex syscall "hang".
(aside) Note that there are 2nd order (if you will) consequences from cleanup of a
failed rpm install transaction that can/will lead to, e.g., duplicate package
entries in a rpmdb, but that's a different, higeher level, issue. FYI a patch for
rpm + Berkeley DB to use "robust mutexes" to reap stale FUTEX's has existed for
about 1.5 years now as well, but is rather tricky to deploy reliably since a functional
"robust mitex" environment has both kernel & glibc version dependencies, and the
patch is specific to rpm, because of the manner that rpm access a Berkeley DB.
So which case are you: 1) registered-but-abandoned dbenv lock 2) stale FUTEX?
I've seen lots of 1) cases, but have only seen/heard of maybe a handful of stale FUTEX's
personally, they appear to be quite rare. But YMMV, everyone's does.
Created attachment 308673 [details]
i catch it again. so, i get rpmdb_stat, but can't understand which numbers i
need to check. see attach: #4
For context (note that you are using an older version than 4.7.25 of Berkeley DB):
The db_stat output is described here:
The locking subsystem is here:
The code (or similar) that is displaying the info is available here, see __lock_printlock():
This looks like a stale futex to me because
1) There are only READ locks. In normal Concurrent Data Store operation, there's
either a single WRITE'r or ultiple READers. So this is not a "normal" operation.
2) All the nerdy hex information being displayed is identical, which is what I would expect
when a common (and stale) kernel futex is the cause of a "hang".
Disclaimer: I have _NOT_verified whether your output is actually a stale futex (or not), just made
a reasoned guess given the supplied db_stat spew. There may well be other interpretations better
Note that there is a patch to use "robust mutexes" that would clear stale futexes in rpm
available somewhere in this thread:
I continue to "sabotage" Fedora bugs ;-)
I have the same problem with yum (3.2.19-3) but only on the 2 last kernel (22.214.171.124-79.fc9.i386 and 126.96.36.199-45). It's working with an older kernel like 188.8.131.52-55
pread64(10, "\0\0\0\0\1\0\0\0\336:\0\0\337:\0\0\335:\0\0\1\0\346\17\0\7b7c91a"..., 4096, 61726720) = 4096
pread64(10, "\0\0\0\0\1\0\0\0\335:\0\0\336:\0\0\334:\0\0\1\0\346\17\0\7c8a29e"..., 4096, 61722624) = 4096
pread64(10, "\0\0\0\0\1\0\0\0\334:\0\0\335:\0\0\333:\0\0\1\0\346\17\0\7\000799f6"..., 4096, 61718528) = 4096
pread64(10, "\0\0\0\0\1\0\0\0\333:\0\0\334:\0\0\332:\0\0\1\0\346\17\0\0076b0e7e"..., 4096, 61714432) = 4096
futex(0xb7a7b850, FUTEX_WAIT, 2, NULL
I have a similar problem on an other computer (fedora 9 as weel but x86_64)
I only have trouble on the installation of the kernel, not for any other packages (so far):
rt_sigprocmask(SIG_SETMASK, NULL, [CHLD], 8) = 0
rt_sigprocmask(SIG_SETMASK, , NULL, 8) = 0
futex(0x21614e8, FUTEX_WAIT_PRIVATE, 2, NULL
^C <unfinished ...>
[root@localhost pascal]# erreur: %post(kernel-184.108.40.206-79.fc9.x86_64) scriptlet failed, signal 2
Although this time it's not completely stuck, I can install the kernel via rpm if I have patience, it takes about 10min.
*** Bug 450705 has been marked as a duplicate of this bug. ***
*** Bug 485780 has been marked as a duplicate of this bug. ***
This message is a reminder that Fedora 9 is nearing its end of life.
Approximately 30 (thirty) days from now Fedora will stop maintaining
and issuing updates for Fedora 9. It is Fedora's policy to close all
bug reports from releases that are no longer maintained. At that time
this bug will be closed as WONTFIX if it remains open with a Fedora
'version' of '9'.
Package Maintainer: If you wish for this bug to remain open because you
plan to fix it in a currently maintained version, simply change the 'version'
to a later Fedora version prior to Fedora 9's end of life.
Bug Reporter: Thank you for reporting this issue and we are sorry that
we may not be able to fix it before Fedora 9 is end of life. If you
would still like to see this bug fixed and are able to reproduce it
against a later version of Fedora please change the 'version' of this
bug to the applicable version. If you are unable to change the version,
please add a comment here and someone will do it for you.
Although we aim to fix as many bugs as possible during every release's
lifetime, sometimes those efforts are overtaken by events. Often a
more recent Fedora release includes newer upstream software that fixes
bugs or makes them obsolete.
The process we are following is described here:
Fedora 9 changed to end-of-life (EOL) status on 2009-07-10. Fedora 9 is
no longer maintained, which means that it will not receive any further
security or bug fix updates. As a result we are closing this bug.
If you can reproduce this bug against a currently maintained version of
Fedora please feel free to reopen this bug against that version.
Thank you for reporting this bug and we are sorry it could not be fixed.