Re: Slow ops on 1 host

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Hey Istvan,
so, osdop_writefull: write full object operations and what this means: in a
way. It replaces previous data, you can think of it as an atomic truncate
to 0 + write from offset 0.  So it always writes (replaces) entire objects.

Cheers!

On Fri, Feb 18, 2022 at 1:33 AM Szabo, Istvan (Agoda) <
Istvan.Szabo@xxxxxxxxx> wrote:

> What writefull actually means, haven't really found anything in ceph doc
> and just some flowchart which not really clear in google :/
>
> This is when it started and have this kind of messages after:
>
> 2022-02-18T08:47:46.694274+0700 osd.148 (osd.148) 1701 : cluster [WRN]
> slow request osd_op(client.524665199.0:984004907 24.74es0
> 24:72fb53ba:::9213182a-14ba-48ad-bde9-289a1c0c0de8.21813
> 595.2__shadow_.6taCcYVmTAl87Ai0ZggPc-b_el2M4A7_3:head [writefull 0~4194304
> in=4194304b] snapc 0=[] ondisk+write+known_if_redirected e63105) initiated
> 2022-02-18T08:47:15.758941+0700 cur
> rently started
>
> Istvan Szabo
> Senior Infrastructure Engineer
> ---------------------------------------------------
> Agoda Services Co., Ltd.
> e: istvan.szabo@xxxxxxxxx
> ---------------------------------------------------
>
> -----Original Message-----
> From: Anthony D'Atri <anthony.datri@xxxxxxxxx>
> Sent: Friday, February 18, 2022 1:44 PM
> To: Szabo, Istvan (Agoda) <Istvan.Szabo@xxxxxxxxx>
> Subject: Re:  Slow ops on 1 host
>
> Email received from the internet. If in doubt, don't click any link nor
> open any attachment !
> ________________________________
>
> Look at the OSD logs, figure out which OSD.  Correlate with syslog / dmesg
> and smartctl etc.  SATA? If it’s all OSDs on the node likely the HBA.  If
> just one, likely the drive.
>
> > On Feb 17, 2022, at 9:44 PM, Szabo, Istvan (Agoda) <
> Istvan.Szabo@xxxxxxxxx> wrote:
> >
> > Hi,
> >
> > Is there a way to identify what and where is stuck on 1 host that
> causing slow operation?
> > If I shutdown that host where the slow ops is coming, the cluster got
> back to normal operation, if I start back, slow ops coming.
> >
> > I'm planning to purge all osds on that host but I scare a bit the slow
> ops will move to another host.
> >
> > Thank you
> > _______________________________________________
> > ceph-users mailing list -- ceph-users@xxxxxxx To unsubscribe send an
> > email to ceph-users-leave@xxxxxxx
>


-- 

Alvaro Soto Escobar

----------------------------------------------------------
Great people talk about ideas,
average people talk about things,
small people talk ... about other people.
_______________________________________________
ceph-users mailing list -- ceph-users@xxxxxxx
To unsubscribe send an email to ceph-users-leave@xxxxxxx




[Index of Archives]     [Information on CEPH]     [Linux Filesystem Development]     [Ceph Development]     [Ceph Large]     [Ceph Dev]     [Linux USB Development]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]     [xfs]


  Powered by Linux