Re: [OmniOS-discuss] Fragmentation

[Richard Elling]

ZIL pre-allocates at the block level, so think along the lines of 12k or 132k.
 — richard

> On Jun 23, 2017, at 11:30 AM, Günther Alka  wrote:
> 
> hello Richard
> 
> I can follow that the Zil does not add more fragmentation to the free space 
> but is this effect relevant?
> If a ZIL pre-allocates say 4G and the remaining fragmented poolsize for 
> regular writes is 12T
> 
> Gea
> 
> Am 23.06.2017 um 19:30 schrieb Richard Elling:
>> A slog helps fragmentation because the space for ZIL is pre-allocated based 
>> on a prediction of
>> how big the write will be. The pre-allocated space includes a 
>> physical-block-sized chain block for the
>> ZIL. An 8k write can allocate 12k for the ZIL entry that is freed when the 
>> txg commits. Thus, a slog
>> can help decrease free space fragmentation in the pool.
>>  — richard
>> 
>> 
>>> On Jun 23, 2017, at 8:56 AM, Guenther Alka  wrote:
>>> 
>>> A Zil or better dedicated Slog device will not help as this is not a write 
>>> cache but a logdevice. Its only there to commit every written datablock and 
>>> to put it onto stable storage. It is read only after a crash to redo a 
>>> missing committed write.
>>> 
>>> All writes, does not matter if sync or not, are going over the rambased 
>>> write cache (per default up to 4GB). This is flushed from time to time as a 
>>> large sequential write. Writes are fragmented then depending on the 
>>> fragmentation of the free space.
>>> 
>>> Gea
>>> 
>>> 
 To prevent it, a ZIL caching all writes (including sync ones, e.g. nfs) 
 can help. Perhaps a DDR drive (or mirror of these) with battery and flash 
 protection from poweroffs, so it does not wear out like flash would. In 
 this case, how-ever random writes come, ZFS does not have to put them on 
 media asap - so it can do larger writes later. This can also protect SSD 
 arrays from excessive small writes and wear-out, though there a bad(ly 
 sized) ZIL can become a bottleneck.
 
 Hope this helps,
 Jim
 --
>>> ___
>>> OmniOS-discuss mailing list
>>> OmniOS-discuss@lists.omniti.com
>>> http://lists.omniti.com/mailman/listinfo/omnios-discuss
> 
> -- 
> ___
> OmniOS-discuss mailing list
> OmniOS-discuss@lists.omniti.com
> http://lists.omniti.com/mailman/listinfo/omnios-discuss

___
OmniOS-discuss mailing list
OmniOS-discuss@lists.omniti.com
http://lists.omniti.com/mailman/listinfo/omnios-discuss

Re: [OmniOS-discuss] Fragmentation

[Günther Alka]

hello Richard

I can follow that the Zil does not add more fragmentation to the free 
space but is this effect relevant?
If a ZIL pre-allocates say 4G and the remaining fragmented poolsize for 
regular writes is 12T


Gea

Am 23.06.2017 um 19:30 schrieb Richard Elling:

A slog helps fragmentation because the space for ZIL is pre-allocated based on 
a prediction of
how big the write will be. The pre-allocated space includes a 
physical-block-sized chain block for the
ZIL. An 8k write can allocate 12k for the ZIL entry that is freed when the txg 
commits. Thus, a slog
can help decrease free space fragmentation in the pool.
  — richard



On Jun 23, 2017, at 8:56 AM, Guenther Alka  wrote:

A Zil or better dedicated Slog device will not help as this is not a write 
cache but a logdevice. Its only there to commit every written datablock and to 
put it onto stable storage. It is read only after a crash to redo a missing 
committed write.

All writes, does not matter if sync or not, are going over the rambased write 
cache (per default up to 4GB). This is flushed from time to time as a large 
sequential write. Writes are fragmented then depending on the fragmentation of 
the free space.

Gea



To prevent it, a ZIL caching all writes (including sync ones, e.g. nfs) can 
help. Perhaps a DDR drive (or mirror of these) with battery and flash 
protection from poweroffs, so it does not wear out like flash would. In this 
case, how-ever random writes come, ZFS does not have to put them on media asap 
- so it can do larger writes later. This can also protect SSD arrays from 
excessive small writes and wear-out, though there a bad(ly sized) ZIL can 
become a bottleneck.

Hope this helps,
Jim
--

___
OmniOS-discuss mailing list
OmniOS-discuss@lists.omniti.com
http://lists.omniti.com/mailman/listinfo/omnios-discuss


--
___
OmniOS-discuss mailing list
OmniOS-discuss@lists.omniti.com
http://lists.omniti.com/mailman/listinfo/omnios-discuss

Re: [OmniOS-discuss] Fragmentation

[Richard Elling]

A slog helps fragmentation because the space for ZIL is pre-allocated based on 
a prediction of
how big the write will be. The pre-allocated space includes a 
physical-block-sized chain block for the
ZIL. An 8k write can allocate 12k for the ZIL entry that is freed when the txg 
commits. Thus, a slog
can help decrease free space fragmentation in the pool.
 — richard


> On Jun 23, 2017, at 8:56 AM, Guenther Alka  wrote:
> 
> A Zil or better dedicated Slog device will not help as this is not a write 
> cache but a logdevice. Its only there to commit every written datablock and 
> to put it onto stable storage. It is read only after a crash to redo a 
> missing committed write.
> 
> All writes, does not matter if sync or not, are going over the rambased write 
> cache (per default up to 4GB). This is flushed from time to time as a large 
> sequential write. Writes are fragmented then depending on the fragmentation 
> of the free space.
> 
> Gea
> 
> 
>> To prevent it, a ZIL caching all writes (including sync ones, e.g. nfs) can 
>> help. Perhaps a DDR drive (or mirror of these) with battery and flash 
>> protection from poweroffs, so it does not wear out like flash would. In this 
>> case, how-ever random writes come, ZFS does not have to put them on media 
>> asap - so it can do larger writes later. This can also protect SSD arrays 
>> from excessive small writes and wear-out, though there a bad(ly sized) ZIL 
>> can become a bottleneck.
>> 
>> Hope this helps,
>> Jim
>> --
> 
> ___
> OmniOS-discuss mailing list
> OmniOS-discuss@lists.omniti.com
> http://lists.omniti.com/mailman/listinfo/omnios-discuss

___
OmniOS-discuss mailing list
OmniOS-discuss@lists.omniti.com
http://lists.omniti.com/mailman/listinfo/omnios-discuss

Re: [OmniOS-discuss] Fragmentation

[Jim Klimov]

On June 23, 2017 4:13:52 PM GMT+02:00, Artyom Zhandarovsky  
wrote:
>disk errors: none
>
>
>
>
>
>-
>
>CAP Alert
>
>-
>
>
>
> Is there any way to decrease fragmentation of dr_tank ?
>
>--
>
>zpool list (Sum of RAW disk capacity without redundancy counted)
>
>--
>
>NAME  SIZE  ALLOC   FREE  EXPANDSZ   FRAGCAP  DEDUP  HEALTH 
>ALTROOT
>
>dr_slow  9.06T  77.6M  9.06T - 0% 0%  1.00x  ONLINE  -
>
>dr_tank  48.9T  35.1T  13.9T -23%71%  1.00x  ONLINE  -
>
>rpool 272G  42.1G   230G -10%15%  1.00x  ONLINE  -
>
>
>
>Real Pool capacity from zfs list
>
>--
>
>NAME   USED AVAILMOUNTPOINT  %
>
>dr_slow   7.69T 1.26T /dr_slow 14%!
>
>dr_tank 41.6T 6.33T /dr_tank 13%!
>
>rpool 45.6G218G  /rpool   83%

The issue of zfs fragmentation is that at some point it becomes hard to find 
free spots to write into, as well as to do large writes contiguously, so 
performance suddenly and noticeably drops. This can impact reads as well, 
especially if atime=on is left as default.

To recover from existing fragmentation you must free up space, perhaps zfs-send 
datasets to another pool, empty as much as you can on this one, and send data 
back - so it lands in large contiguous writes.

To prevent it, a ZIL caching all writes (including sync ones, e.g. nfs) can 
help. Perhaps a DDR drive (or mirror of these) with battery and flash 
protection from poweroffs, so it does not wear out like flash would. In this 
case, how-ever random writes come, ZFS does not have to put them on media asap 
- so it can do larger writes later. This can also protect SSD arrays from 
excessive small writes and wear-out, though there a bad(ly sized) ZIL can 
become a bottleneck.

Hope this helps,
Jim
--
Typos courtesy of K-9 Mail on my Redmi Android
___
OmniOS-discuss mailing list
OmniOS-discuss@lists.omniti.com
http://lists.omniti.com/mailman/listinfo/omnios-discuss

Re: [OmniOS-discuss] Loosing NFS shares

[Guenther Alka]

With ESXi 6.0 I have had NFS problems as well. You should use at least 
6.0U2(or 6.5.0d with the 6.5 line)


Another problem may be timeouts. ZFS will wait longer for a disk than 
ESXi for NFS. What you can do is reducing disk timeout time in 
/etc/system with set sd:sd_io_time=0xF (=15s, default is 60s).  Check 
system-logs for disk related problems (/var/adm/messages).


You should also use vmxnet3s for OmniOS with increased buffer settings 
for vmxnet3, tcp and NFS for 10G networking.Snaps and replication itself 
are uncritical as long as you do not have disk or network problems. Ex 
MTU 9000 can give problems. Enough RAM may be another item. You should 
give 8GB or more to OmniOS.



Gea
napp-it


Am 23.06.2017 um 08:59 schrieb Oliver Weinmann:


Hi,

We have a hyperconverged setup.

ESXi 6.0 -> OmniOS VM -> Storage Passthrough. The 10GB Nics are in 
configured in Active / Standby. For NFS I use a dedicated /24 VLAN.


I wonder if ZFS replication or snapshotting could be the reason for 
the problems we are seeing. The system fails at night time. We have 
several auto-sync jobs from a Nexenta system to this system. I have 
now disabled these jobs. I just wonder where logs etc. I can find any 
information on what is going on on the system when it fails.


The symtoms are always the same. The vmware nfsshare disappear. So I 
just reshare them (nfs set nfsshare=…. /hgst4u60/vmware-ds-1. Then the 
root ZFS folders of all local drives that I think have snapshots 
enabled are unmounted. So I have to unmount them all hard and remount 
them. Also the root ZFS folders from replicated ZFS folders are 
unmounted. This is why I assume that there is a link between the 
problems and auto-sync or snapshotting.


*Oliver Weinmann*
Senior Unix VMWare, Storage Engineer

Telespazio VEGA Deutschland GmbH
Europaplatz 5 - 64293 Darmstadt - Germany
Ph: + 49 (0)6151 8257 744 | Fax: +49 (0)6151 8257 799
oliver.weinm...@telespazio-vega.de 


http://www.telespazio-vega.de

Registered office/Sitz: Darmstadt, Register court/Registergericht: 
Darmstadt, HRB 89231; Managing Director/Geschäftsführer: Sigmar Keller


*From:*Lawrence Giam [mailto:paladinemisha...@gmail.com]
*Sent:* Freitag, 23. Juni 2017 05:00
*To:* Oliver Weinmann 
*Subject:* Re: [OmniOS-discuss] Loosing NFS shares

Hi Oliver,

What is your network setup in relationship to NFS file server and VM 
Host? Is your backdone using 10Gbe network and are they setup with 
redundancy (eg. MLAG)?


Regards.

On Thu, Jun 22, 2017 at 2:45 PM, Oliver Weinmann 
> wrote:


Hi,

we are using OmniOS for a few months now and have big trouble with
stability. We mainly use it for VMware NFS datastores. The last 3
nights we lost all NFS datastores and VMs stopped running. I
noticed that even though zfs get sharenfs shows folders as shared
they become inaccessible. Setting sharenfs to off and sharing
again solves the issue. I have no clue where to start. I’m fairly
new to OmniOS.

Any help would be highly appreciated.

Thanks and Best Regards,

Oliver

*Oliver Weinmann*
Senior Unix VMWare, Storage Engineer

Telespazio VEGA Deutschland GmbH
Europaplatz 5 - 64293 Darmstadt - Germany
Ph: + 49 (0)6151 8257 744  | Fax: +49
(0)6151 8257 799 
oliver.weinm...@telespazio-vega.de

http://www.telespazio-vega.de

Registered office/Sitz: Darmstadt, Register court/Registergericht:
Darmstadt, HRB 89231; Managing Director/Geschäftsführer: Sigmar Keller


___
OmniOS-discuss mailing list
OmniOS-discuss@lists.omniti.com

http://lists.omniti.com/mailman/listinfo/omnios-discuss



___
OmniOS-discuss mailing list
OmniOS-discuss@lists.omniti.com
http://lists.omniti.com/mailman/listinfo/omnios-discuss


--

___
OmniOS-discuss mailing list
OmniOS-discuss@lists.omniti.com
http://lists.omniti.com/mailman/listinfo/omnios-discuss