https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Bug ID: 1064929 Summary: Kernel container scalability issue in docker devicemapper backend Product: Red Hat Enterprise Linux 7 Version: 7.0 Component: kernel Severity: high Assignee: kernel-mgr@redhat.com Reporter: sct@redhat.com QA Contact: kernel-qe@redhat.com CC: agk@redhat.com, alexl@redhat.com, golang@lists.fedoraproject.org, jeder@redhat.com, lsm5@redhat.com, mattdm@redhat.com, mgoldman@redhat.com, skottler@redhat.com, vbatts@redhat.com Depends On: 1061359 Blocks: 1064012 Group: private
Referenced Bugs:
https://bugzilla.redhat.com/show_bug.cgi?id=1061359 [Bug 1061359] improve docker devicemapper backend scalability https://bugzilla.redhat.com/show_bug.cgi?id=1064012 [Bug 1064012] improve docker devicemapper backend scalability
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Stephen Tweedie sct@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Blocks|1064012 |
--- Comment #2 from Stephen Tweedie sct@redhat.com --- *** Bug 1064012 has been marked as a duplicate of this bug. ***
Referenced Bugs:
https://bugzilla.redhat.com/show_bug.cgi?id=1064012 [Bug 1064012] improve docker devicemapper backend scalability
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Jeremy Eder jeder@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Priority|unspecified |high CC| |perfbz@redhat.com
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Jeremy Eder jeder@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Priority|high |urgent Severity|high |urgent
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Linda Wang lwang@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Assignee|kernel-mgr@redhat.com |aviro@redhat.com
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Ric Wheeler rwheeler@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Summary|Kernel container |Kernel container |scalability issue in docker |scalability issue in docker |devicemapper backend |- VFS and mount path | |consuming too much CPU
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Ian Kent ikent@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |ikent@redhat.com
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
--- Comment #12 from Ian Kent ikent@redhat.com --- Created attachment 865491 --> https://bugzilla.redhat.com/attachment.cgi?id=865491&action=edit kernel config diff
I was looking mostly at the "Lock Debugging (spinlocks, mutexes, etc...)" and the "RCU Debugging" sections but didn't see what I thought were similar differences between the rhel7 and ubuntu configs. on the test machine.
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
--- Comment #17 from Stephen Tweedie sct@redhat.com --- Created attachment 865585 --> https://bugzilla.redhat.com/attachment.cgi?id=865585&action=edit Disable CONFIG_SLUB_DEBUG
Patch against dist-git kernel rhel-7.0 branch to disable CONFIG_SLUB_DEBUG.
The -BuildRequires: rpm-build >= 4.9.0-1, elfutils >= elfutils-0.153-1 +BuildRequires: rpm-build >= 4.9.0-1, elfutils >= 0.153-1 is needed to satisfy mock (elfutils-0.153-1 has two "-" separators and is rejected as an invalid VR.)
Build with $ rhpkg srpm $ mock -r rhel-7-x86_64 --without=debug --without=debuginfo $PATH_TO_SRPM
NB -- the patch updates ONLY the kernel-x86_64 config, not kernel-debug nor other architectures.
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Stephen Tweedie sct@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Blocks| |1069718
Referenced Bugs:
https://bugzilla.redhat.com/show_bug.cgi?id=1069718 [Bug 1069718] Kernel container scalability issue in docker - VFS and mount path consuming too much CPU
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Bhavna Sarathy bsarathy@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Blocks| |1069814
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
--- Comment #23 from Stephen Tweedie sct@redhat.com --- Created attachment 867566 --> https://bugzilla.redhat.com/attachment.cgi?id=867566&action=edit Debug info from docker containers
Tarball containing:
-rw-r--r-- sct/sct 8428 2014-02-24 19:34 mountinfo.1.txt
/proc/1/mountinfo root namespace mount info (after 3 docker containers created)
-rw-r--r-- sct/sct 1485 2014-02-24 19:34 mountinfo.d1.txt -rw-r--r-- sct/sct 1485 2014-02-24 19:34 mountinfo.d2.txt -rw-r--r-- sct/sct 1485 2014-02-24 19:34 mountinfo.d3.txt
/proc/$pid/mountinfo from those 3 containers; and
-rw-r--r-- sct/sct 332960 2014-02-24 19:41 docker.strace
strace -f of the docker daemon creating a single container, including all the various mount and unshare calls.
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Stephen Tweedie sct@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Blocks| |1072446
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Stephen Tweedie sct@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Blocks| |1072451
Referenced Bugs:
https://bugzilla.redhat.com/show_bug.cgi?id=1072451 [Bug 1072451] Kernel container scalability issue in docker - [2/4] Expand size of mount hash table
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Stephen Tweedie sct@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Blocks| |1072457
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Stephen Tweedie sct@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Blocks| |1072461
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Paul W. Frields pfrields@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Keywords| |Tracking Status|NEW |MODIFIED Blocks|1072446, 1072451, 1072457, | |1072461 | Depends On| |1072446, 1072451, 1072457, | |1072461
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Josh Poimboeuf jpoimboe@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC|jpoimboe@redhat.com |
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Linqing Lu lilu@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |lilu@redhat.com QA Contact|kernel-qe@redhat.com |fs-qe@redhat.com
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Eryu Guan eguan@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- CC| |eguan@redhat.com QA Contact|fs-qe@redhat.com |eguan@redhat.com
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Eryu Guan eguan@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|MODIFIED |VERIFIED
--- Comment #29 from Eryu Guan eguan@redhat.com --- Reproduced with kernel-3.10.0-106.el7, following steps in comment 5
Samples: 931K of event 'cycles', Event count (approx.): 39097077164 37.59% [kernel] [k] pcpu_alloc_area 17.67% [kernel] [k] pcpu_free_area 7.97% [kernel] [k] lg_global_lock 6.90% [kernel] [k] seq_list_start 3.57% [kernel] [k] memset 3.52% [kernel] [k] find_next_bit 3.02% [kernel] [k] lg_global_unlock 1.69% libc-2.17.so [.] _IO_vfscanf
Verified with kernel-3.10.0-121.el7
Samples: 253K of event 'cycles', Event count (approx.): 70681567890 13.92% libc-2.17.so [.] _IO_vfscanf 6.75% systemd [.] 0x0000000000080922 5.26% libc-2.17.so [.] malloc_consolidate 5.14% libc-2.17.so [.] _int_malloc 4.77% [kernel] [k] strchr 2.72% libc-2.17.so [.] _int_free 2.50% [kernel] [k] memcpy 2.25% [kernel] [k] format_decode 2.05% libmount.so.1.1.0 [.] 0x00000000000149b1 1.87% [kernel] [k] seq_escape 1.80% [kernel] [k] vsnprintf 1.74% libc-2.17.so [.] __strcmp_sse42
Also autofs regression/nfs regression tests passed without hitting new issues.
(Actually there's one regression in autofs tests, see bug 1083815, but that's probably caused by bug 1031475)
QE also have done three rounds testings since Snapshot 12 (on snapshot 13, RC-1 and RC-2), no other severe issues found.
So set to VERIFIED.
https://bugzilla.redhat.com/show_bug.cgi?id=1064929
Ludek Smid lsmid@redhat.com changed:
What |Removed |Added ---------------------------------------------------------------------------- Status|VERIFIED |CLOSED Resolution|--- |CURRENTRELEASE Last Closed| |2014-06-26 07:36:03
--- Comment #30 from Ludek Smid lsmid@redhat.com --- This request was resolved in Red Hat Enterprise Linux 7.0.
Contact your manager or support representative in case you have further questions about the request.
https://bugzilla.redhat.com/show_bug.cgi?id=1064929 Bug 1064929 depends on bug 1061359, which changed state.
Bug 1061359 Summary: improve docker devicemapper backend scalability https://bugzilla.redhat.com/show_bug.cgi?id=1061359
What |Removed |Added ---------------------------------------------------------------------------- Status|ASSIGNED |CLOSED Resolution|--- |NOTABUG
golang@lists.fedoraproject.org