Date: Wed, 15 Jul 2015 12:36:13 +1000 From: Jan Mikkelsen <janm@transactionware.com> To: Karl Denninger <karl@denninger.net> Cc: freebsd-fs@freebsd.org Subject: Re: FreeBSD 10.1 Memory Exhaustion Message-ID: <B3777FBA-EF15-49E3-8FCE-468955EF2ED6@transactionware.com> In-Reply-To: <55A3A800.5060904@denninger.net> References: <CAB2_NwCngPqFH4q-YZk00RO_aVF9JraeSsVX3xS0z5EV3YGa1Q@mail.gmail.com> <55A3A800.5060904@denninger.net>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi,
To make the 2015-02-05 patch work on 10.2-BETA1 I had to add this patch =
as well:
--- =
//depot/vendor/freebsd/10-stable-local/src/sys/cddl/contrib/opensolaris/ut=
s/common/fs/zfs/zio.c 2015-07-03 02:03:52.000000000 1000
+++ =
/home/janm/p4/freebsd-std/FreeBSD/src/sys/cddl/contrib/opensolaris/uts/com=
mon/fs/zfs/zio.c 2015-07-03 02:03:52.000000000 1000
@@ -44,9 +44,9 @@
SYSCTL_DECL(_vfs_zfs);
SYSCTL_NODE(_vfs_zfs, OID_AUTO, zio, CTLFLAG_RW, 0, "ZFS ZIO");
#if defined(__amd64__)
-static int zio_use_uma =3D 1;
+int zio_use_uma =3D 1;
#else
-static int zio_use_uma =3D 0;
+int zio_use_uma =3D 0;
#endif
TUNABLE_INT("vfs.zfs.zio.use_uma", &zio_use_uma);
SYSCTL_INT(_vfs_zfs_zio, OID_AUTO, use_uma, CTLFLAG_RDTUN, =
&zio_use_uma, 0,
Without changing the storage class of zio_use_uma this error appeared at =
runtime "link_elf: symbol zio_use_uma undefined=94. This is unsurprising =
because the linker can=92t resolve an extern symbol to a static in =
another file. How are other people using this patch, or am I missing =
something fundamental?
Regards,
Jan.
> On 13 Jul 2015, at 21:58, Karl Denninger <karl@denninger.net> wrote:
>=20
> Put this on your box and see if the problem goes away.... :-)
>=20
> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D187594
>=20
> The 2015-02-10 refactor will apply against 10.1-STABLE and 10.2-PRE =
(the
> latter will give you a 10-line fuzz in one block but applies and =
works.)
>=20
> I've been unable to provoke misbehavior with this patch in and I run a
> cron job that does auto-snapshotting. There are others that have run
> this patch with similarly positive results.
>=20
> On 7/13/2015 06:48, Christopher Forgeron wrote:
>> TL;DR Summary: I can run FreeBSD out of memory quite consistently, =
and it=92s
>> not a TOS/mbuf exhaustion issue. It=92s quite possible that ZFS is =
the
>> culprit, but shouldn=92t the pager be able to handle aggressive =
memory
>> requests in a low memory situation gracefully, without needing custom
>> tuning of ZFS / VM?
>>=20
>>=20
>> Hello,
>>=20
>> I=92ve been dealing with some instability in my 10.1-RELEASE and
>> STABLEr282701M machines for the last few months.
>>=20
>> These machines are NFS/iSCSI storage machines, running on Dell M610x =
or
>> similar hardware, 96 Gig Memory, 10Gig Network Cards, dual Xeon =
Processors
>> =96 Fairly beefy stuff.
>>=20
>> Initially I thought it was more issues with TOS / jumbo mbufs, as I =
had
>> this problem last year. I had thought that this was properly =
resolved, but
>> setting my MTU to 1500, and turning off TOS did give me a bit more
>> stability. Currently all my machines are set this way.
>>=20
>> Crashes were usually represented by loss of network connectivity, and =
the
>> ctld daemon scrolling messages across the screen at full speed about =
lost
>> connections.
>>=20
>> All of this did seem like more network stack problems, but with each =
crash
>> I=92d be able to learn a bit more.
>>=20
>> Usually there was nothing of any use in the logfile, but every now =
and then
>> I=92d get this:
>>=20
>> Jun 3 13:02:04 san0 kernel: WARNING: 172.16.0.97
>> (iqn.1998-01.com.vmware:esx5a-3387a188): failed to allocate memory
>> Jun 3 13:02:04 san0 kernel: WARNING: icl_pdu_new: failed to allocate =
80
>> bytes
>> Jun 3 13:02:04 san0 kernel: WARNING: 172.16.0.97
>> (iqn.1998-01.com.vmware:esx5a-3387a188): failed to allocate memory
>> Jun 3 13:02:04 san0 kernel: WARNING: icl_pdu_new: failed to allocate =
80
>> bytes
>> Jun 3 13:02:04 san0 kernel: WARNING: 172.16.0.97
>> (iqn.1998-01.com.vmware:esx5a-3387a188): failed to allocate memory
>> ---------
>> Jun 4 03:03:09 san0 kernel: WARNING: icl_pdu_new: failed to allocate =
80
>> bytes
>> Jun 4 03:03:09 san0 kernel: WARNING: icl_pdu_new: failed to allocate =
80
>> bytes
>> Jun 4 03:03:09 san0 kernel: WARNING: 172.16.0.97
>> (iqn.1998-01.com.vmware:esx5a-3387a188): failed to allocate memory
>> Jun 4 03:03:09 san0 kernel: WARNING: 172.16.0.97
>> (iqn.1998-01.com.vmware:esx5a-3387a188): connection error; dropping
>> connection
>> Jun 4 03:03:09 san0 kernel: WARNING: 172.16.0.97
>> (iqn.1998-01.com.vmware:esx5a-3387a188): connection error; dropping
>> connection
>> Jun 4 03:03:10 san0 kernel: WARNING: 172.16.0.97
>> (iqn.1998-01.com.vmware:esx5a-3387a188): waiting for CTL to terminate
>> tasks, 1 remaining
>> Jun 4 06:04:27 san0 syslogd: kernel boot file is /boot/kernel/kernel
>>=20
>> So knowing that it seemed to be running out of memory, I started =
leaving
>> leaving =91vmstat 5=92 running on a console, to see what it was =
displaying
>> during the crash.
>>=20
>> It was always the same thing:
>>=20
>> 0 0 0 1520M 4408M 15 0 0 0 25 19 0 0 21962 1667 =
91390
>> 0 33 67
>> 0 0 0 1520M 4310M 9 0 0 0 2 15 3 0 21527 1385 =
95165
>> 0 31 69
>> 0 0 0 1520M 4254M 7 0 0 0 14 19 0 0 17664 1739 =
72873
>> 0 18 82
>> 0 0 0 1520M 4145M 2 0 0 0 0 19 0 0 23557 1447 =
96941
>> 0 36 64
>> 0 0 0 1520M 4013M 4 0 0 0 14 19 0 0 4288 490 =
34685
>> 0 72 28
>> 0 0 0 1520M 3885M 2 0 0 0 0 19 0 0 11141 1038 =
69242
>> 0 52 48
>> 0 0 0 1520M 3803M 10 0 0 0 14 19 0 0 24102 1834 =
91050
>> 0 33 67
>> 0 0 0 1520M 8192B 2 0 0 0 2 15 1 0 19037 1131 =
77470
>> 0 45 55
>> 0 0 0 1520M 8192B 0 22 0 0 2 0 6 0 146 82 =
578 0
>> 0 100
>> 0 0 0 1520M 8192B 1 0 0 0 0 0 0 0 130 40 =
510 0
>> 0 100
>> 0 0 0 1520M 8192B 0 0 0 0 0 0 0 0 143 40 =
501 0
>> 0 100
>> 0 0 0 1520M 8192B 0 0 0 0 0 0 0 0 201 62 =
660 0
>> 0 100
>> 0 0 0 1520M 8192B 0 0 0 0 0 0 0 0 101 28 =
404 0
>> 0 100
>> 0 0 0 1520M 8192B 0 0 0 0 0 0 0 0 97 27 =
398 0
>> 0 100
>> 0 0 0 1520M 8192B 0 0 0 0 0 0 0 0 93 28 =
377 0
>> 0 100
>> 0 0 0 1520M 8192B 0 0 0 0 0 0 0 0 92 27 =
373 0
>> 0 100
>>=20
>>=20
>> I=92d go from a decent amount of free memory to suddenly having none. =
Vmstat
>> would stop outputting, console commands would hang, etc. The whole =
system
>> would be useless.
>>=20
>> Looking into this, I came across a similar issue;
>>=20
>> https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D199189
>>=20
>> I started increasing v.v_free_min, and it helped =96 My crashes went =
from
>> being ~every 6 hours to every few days.
>>=20
>> Currently I=92m running with vm.v_free_min=3D1254507 =96 That=92s =
(1254507 * 4KiB)
>> , or 4.78GiB of Reserve. The vmstat above is of a machine with that
>> setting still running to 8B of memory.
>>=20
>> I have two issues here:
>>=20
>> 1) I don=92t think I should ever be able to run the system into the =
ground on
>> memory. Deny me new memory until the pager can free more.
>> 2) Setting =91min=92 doesn=92t really mean =91min=92 as it can =
obviously go below
>> that threshold.
>>=20
>>=20
>> I have plenty of local UFS swap (non-ZFS drives)
>>=20
>> Adrian requested that I output a few more diagnostic items, and this =
is
>> what I=92m running on a console now, in a loop:
>>=20
>> vmstat
>> netstat -m
>> vmstat -z
>> sleep 1
>>=20
>> The output of four crashes are attached here, as they can be a bit =
long.
>> Let me know if that=92s not a good way to report them. They will each =
start
>> mid-way through a vmstat =96z output, as that=92s as far back as my =
terminal
>> buffer allows.
>>=20
>>=20
>>=20
>> Now, I have a good idea of the conditions that are causing this: ZFS
>> Snapshots, run by cron, during times of high ZFS writes.
>>=20
>> The crashes are all nearly on the hour, as that=92s when crontab =
triggers my
>> python scripts to make new snapshots, and delete old ones.
>>=20
>> My average FreeBSD machine has ~ 30 zfs datasets, with each pool =
having ~20
>> TiB used. These all need to snapshot on the hour.
>>=20
>> By staggering the snapshots by a few minutes, I have been able to =
reduce
>> crashing from every other day to perhaps once a week if I=92m lucky =96=
But if
>> I start moving a lot of data around, I can cause daily crashes again.
>>=20
>> It=92s looking to be the memory demand of snapshotting lots of ZFS =
datasets
>> at the same time while accepting a lot of write traffic.
>>=20
>> Now perhaps the answer is =91don=92t do that=92 but I feel that =
FreeBSD should be
>> robust enough to handle this. I don=92t mind tuning for now to
>> reduce/eliminate this, but others shouldn=92t run into this pain just =
because
>> they heavily load their machines =96 There must be a way of avoiding =
this
>> condition.
>>=20
>> Here are the contents of my /boot/loader.conf and sysctl.conf, so =
show my
>> minimal tuning to make this problem a little more bearable:
>>=20
>> /boot/loader.conf
>> vfs.zfs.arc_meta_limit=3D49656727553
>> vfs.zfs.arc_max =3D 91489280512
>>=20
>> /etc/sysctl.conf
>> vm.v_free_min=3D1254507
>>=20
>>=20
>> Any suggestions/help is appreciated.
>>=20
>> Thank you.
>>=20
>>=20
>> _______________________________________________
>> freebsd-fs@freebsd.org mailing list
>> http://lists.freebsd.org/mailman/listinfo/freebsd-fs
>> To unsubscribe, send any mail to "freebsd-fs-unsubscribe@freebsd.org"
>=20
> --=20
> Karl Denninger
> karl@denninger.net <mailto:karl@denninger.net>
> /The Market Ticker/
> /[S/MIME encrypted email preferred]/
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?B3777FBA-EF15-49E3-8FCE-468955EF2ED6>