Regular errors on ceph pg's!

alexskysilk said:
The biggest issue with a raid controller masquerading as JBOD is the inconsistent queue depth. Your disks are designed for 128-256, but your "raid volume" is (probably) set to 512 or more.
Click to expand...

These controllers have the ability to set this parameter. Now the parameter is set to Auto. Change to 256? Or 128?

alexskysilk said:
If you really insist of using this hardware, you REALLY REALLY want an SSD for your db and wal traffic.
Click to expand...

I understand this, but there is no such possibility, unfortunately. :(
 
Again....

Code: 
2018-06-19 14:00:00.000210 mon.cn1 mon.0 192.168.110.1:6789/0 11002 : cluster [INF] overall HEALTH_OK
2018-06-19 15:00:00.000267 mon.cn1 mon.0 192.168.110.1:6789/0 11294 : cluster [INF] overall HEALTH_OK
2018-06-19 16:00:00.000298 mon.cn1 mon.0 192.168.110.1:6789/0 11573 : cluster [INF] overall HEALTH_OK
2018-06-19 17:00:00.000172 mon.cn1 mon.0 192.168.110.1:6789/0 11855 : cluster [INF] overall HEALTH_OK
2018-06-19 18:00:00.000304 mon.cn1 mon.0 192.168.110.1:6789/0 12078 : cluster [INF] overall HEALTH_OK
2018-06-19 19:00:00.000166 mon.cn1 mon.0 192.168.110.1:6789/0 12329 : cluster [INF] overall HEALTH_OK
2018-06-19 20:00:00.000235 mon.cn1 mon.0 192.168.110.1:6789/0 12551 : cluster [INF] overall HEALTH_OK
2018-06-19 20:33:13.751966 osd.21 osd.21 192.168.110.4:6800/2548 43 : cluster [ERR] 4.353 shard 21: soid 4:cac452b8:::rbd_data.18c6c2ae8944a.0000000000001025:head candidate had a read error
2018-06-19 20:33:55.410412 osd.21 osd.21 192.168.110.4:6800/2548 44 : cluster [ERR] 4.353 shard 21: soid 4:cae217dc:::rbd_data.1eb842ae8944a.0000000000006e51:head candidate had a read error
2018-06-19 20:34:01.129939 osd.21 osd.21 192.168.110.4:6800/2548 45 : cluster [ERR] 4.353 shard 21: soid 4:cae71552:::rbd_data.266182ae8944a.0000000000003c88:head candidate had a read error
2018-06-19 20:34:37.841588 mon.cn1 mon.0 192.168.110.1:6789/0 12714 : cluster [ERR] Health check failed: 3 scrub errors (OSD_SCRUB_ERRORS)
2018-06-19 20:34:37.841732 mon.cn1 mon.0 192.168.110.1:6789/0 12715 : cluster [ERR] Health check failed: Possible data damage: 1 pg inconsistent (PG_DAMAGED)
2018-06-19 20:34:33.707308 osd.21 osd.21 192.168.110.4:6800/2548 46 : cluster [ERR] 4.353 deep-scrub 0 missing, 3 inconsistent objects
2018-06-19 20:34:33.707315 osd.21 osd.21 192.168.110.4:6800/2548 47 : cluster [ERR] 4.353 deep-scrub 3 errors
2018-06-19 20:37:54.419003 osd.8 osd.8 192.168.110.2:6808/3144 33 : cluster [ERR] 4.2a7 shard 21: soid 4:e56bbb6c:::rbd_data.b824874b0dc51.000000000000662c:head candidate had a read error
2018-06-19 20:38:13.614101 osd.8 osd.8 192.168.110.2:6808/3144 34 : cluster [ERR] 4.2a7 deep-scrub 0 missing, 1 inconsistent objects
2018-06-19 20:38:13.614108 osd.8 osd.8 192.168.110.2:6808/3144 35 : cluster [ERR] 4.2a7 deep-scrub 1 errors
2018-06-19 20:38:16.141210 mon.cn1 mon.0 192.168.110.1:6789/0 12743 : cluster [ERR] Health check update: 4 scrub errors (OSD_SCRUB_ERRORS)
2018-06-19 20:38:16.141348 mon.cn1 mon.0 192.168.110.1:6789/0 12744 : cluster [ERR] Health check update: Possible data damage: 2 pgs inconsistent (PG_DAMAGED)
2018-06-19 20:51:28.594242 mon.cn1 mon.0 192.168.110.1:6789/0 12838 : cluster [ERR] Health check update: Possible data damage: 2 pgs inconsistent, 1 pg repair (PG_DAMAGED)
2018-06-19 20:52:43.377943 mon.cn1 mon.0 192.168.110.1:6789/0 12846 : cluster [ERR] Health check update: 1 scrub errors (OSD_SCRUB_ERRORS)
2018-06-19 20:52:43.378028 mon.cn1 mon.0 192.168.110.1:6789/0 12847 : cluster [ERR] Health check update: Possible data damage: 1 pg inconsistent (PG_DAMAGED)
2018-06-19 20:55:35.404671 mon.cn1 mon.0 192.168.110.1:6789/0 12866 : cluster [ERR] Health check update: Possible data damage: 1 pg inconsistent, 1 pg repair (PG_DAMAGED)
2018-06-19 20:56:45.904782 mon.cn1 mon.0 192.168.110.1:6789/0 12872 : cluster [INF] Health check cleared: OSD_SCRUB_ERRORS (was: 1 scrub errors)
2018-06-19 20:56:45.904884 mon.cn1 mon.0 192.168.110.1:6789/0 12873 : cluster [INF] Health check cleared: PG_DAMAGED (was: Possible data damage: 1 pg inconsistent, 1 pg repair)
2018-06-19 20:56:45.904924 mon.cn1 mon.0 192.168.110.1:6789/0 12874 : cluster [INF] Cluster is now healthy
2018-06-19 21:00:00.000132 mon.cn1 mon.0 192.168.110.1:6789/0 12885 : cluster [INF] overall HEALTH_OK
2018-06-19 22:00:00.000309 mon.cn1 mon.0 192.168.110.1:6789/0 13308 : cluster [INF] overall HEALTH_OK
2018-06-19 22:06:32.092230 mon.cn1 mon.0 192.168.110.1:6789/0 13366 : cluster [INF] osd.21 failed (root=default,host=cn4) (connection refused reported by osd.0)
2018-06-19 22:06:32.153206 mon.cn1 mon.0 192.168.110.1:6789/0 13389 : cluster [WRN] Health check failed: 1 osds down (OSD_DOWN)
2018-06-19 22:06:33.294838 mon.cn1 mon.0 192.168.110.1:6789/0 13392 : cluster [WRN] Health check failed: Reduced data availability: 6 pgs peering (PG_AVAILABILITY)
2018-06-19 22:06:33.294926 mon.cn1 mon.0 192.168.110.1:6789/0 13393 : cluster [WRN] Health check failed: Degraded data redundancy: 18864/2230887 objects degraded (0.846%), 26 pgs degraded (PG_DEGRADED)
2018-06-19 22:06:39.150868 mon.cn1 mon.0 192.168.110.1:6789/0 13397 : cluster [WRN] Health check update: Degraded data redundancy: 189336/2230887 objects degraded (8.487%), 261 pgs degraded (PG_DEGRADED)
2018-06-19 22:06:39.150932 mon.cn1 mon.0 192.168.110.1:6789/0 13398 : cluster [INF] Health check cleared: PG_AVAILABILITY (was: Reduced data availability: 6 pgs peering)
2018-06-19 22:07:30.752171 mon.cn1 mon.0 192.168.110.1:6789/0 13410 : cluster [INF] Health check cleared: OSD_DOWN (was: 1 osds down)
2018-06-19 22:07:30.758541 mon.cn1 mon.0 192.168.110.1:6789/0 13411 : cluster [INF] osd.21 192.168.110.4:6800/470611 boot
2018-06-19 22:07:31.762967 mon.cn1 mon.0 192.168.110.1:6789/0 13413 : cluster [WRN] Health check update: Degraded data redundancy: 173192/2230887 objects degraded (7.763%), 239 pgs degraded (PG_DEGRADED)
2018-06-19 22:07:33.773546 mon.cn1 mon.0 192.168.110.1:6789/0 13416 : cluster [WRN] Health check failed: Reduced data availability: 10 pgs peering (PG_AVAILABILITY)
2018-06-19 22:07:37.830541 mon.cn1 mon.0 192.168.110.1:6789/0 13418 : cluster [WRN] Health check update: Degraded data redundancy: 101/2230887 objects degraded (0.005%), 82 pgs degraded (PG_DEGRADED)
2018-06-19 22:07:37.830702 mon.cn1 mon.0 192.168.110.1:6789/0 13419 : cluster [INF] Health check cleared: PG_AVAILABILITY (was: Reduced data availability: 13 pgs peering)
2018-06-19 22:07:44.470322 mon.cn1 mon.0 192.168.110.1:6789/0 13420 : cluster [WRN] Health check update: Degraded data redundancy: 70/2230887 objects degraded (0.003%), 59 pgs degraded (PG_DEGRADED)
2018-06-19 22:07:49.974957 mon.cn1 mon.0 192.168.110.1:6789/0 13421 : cluster [WRN] Health check update: Degraded data redundancy: 34/2230887 objects degraded (0.002%), 30 pgs degraded (PG_DEGRADED)
2018-06-19 22:07:56.015198 mon.cn1 mon.0 192.168.110.1:6789/0 13422 : cluster [WRN] Health check update: Degraded data redundancy: 12/2230887 objects degraded (0.001%), 10 pgs degraded (PG_DEGRADED)
2018-06-19 22:08:00.094944 mon.cn1 mon.0 192.168.110.1:6789/0 13423 : cluster [INF] Health check cleared: PG_DEGRADED (was: Degraded data redundancy: 12/2230887 objects degraded (0.001%), 10 pgs degraded)
2018-06-19 22:08:00.095030 mon.cn1 mon.0 192.168.110.1:6789/0 13424 : cluster [INF] Cluster is now healthy


It seems that there is no solution to this problem for me.
Only the replacement of disks, but whether it helps, I do not know. And there is no such possibility.
Either return everything to PVE 4.x without Bluestore.
:(
 
if everything was working for you with proxmox 4.4/filestore, you can always regress back, and you can still do filestore with luminous...

that said, I suggest you may want to rethink this solution alltogether- create a RAID10 volume and use it directly.
 
gosha said:
These controllers have the ability to set this parameter. Now the parameter is set to Auto. Change to 256? Or 128?
Click to expand...
I guess, the suggestion from @alexskysilk is targeted towards SSD/NVMe, a spinner has (usually) queue depth of 32.
 
Hi!

Currently I transferred a deep_scrub to non-production hours (Only backup all VMs & CTs to the server placed outside the cluster):

debug ms = 0/0
osd scrub begin hour = 0
osd scrub end hour = 8
osd scrub sleep = 0.1

All day "cluster [INF] overall HEALTH_OK" and errors of deep scrubbing at night disappeared.
But there were Warnings:
Code: 
2018-06-26 00:09:37.179785 mon.cn1 mon.0 192.168.110.1:6789/0 57560 : cluster [WRN] Health check failed: 50 slow requests are blocked > 32 sec (REQUEST_SLOW)
2018-06-26 00:09:44.305499 mon.cn1 mon.0 192.168.110.1:6789/0 57562 : cluster [WRN] Health check update: 49 slow requests are blocked > 32 sec (REQUEST_SLOW)
2018-06-26 00:09:49.305951 mon.cn1 mon.0 192.168.110.1:6789/0 57564 : cluster [WRN] Health check update: 48 slow requests are blocked > 32 sec (REQUEST_SLOW)
2018-06-26 00:09:54.306428 mon.cn1 mon.0 192.168.110.1:6789/0 57566 : cluster [WRN] Health check update: 45 slow requests are blocked > 32 sec (REQUEST_SLOW)
2018-06-26 00:09:59.306810 mon.cn1 mon.0 192.168.110.1:6789/0 57568 : cluster [WRN] Health check update: 42 slow requests are blocked > 32 sec (REQUEST_SLOW)
2018-06-26 00:10:02.376282 mon.cn1 mon.0 192.168.110.1:6789/0 57569 : cluster [INF] Health check cleared: REQUEST_SLOW (was: 42 slow requests are blocked > 32 sec)
2018-06-26 00:10:02.376392 mon.cn1 mon.0 192.168.110.1:6789/0 57570 : cluster [INF] Cluster is now healthy

It looks like the whole problem in low performance HDDs.
In this regard, the question:
Anybody can give recommendations for tuning settings for slow HDDs even if this results in a slightly degraded performance?

Gosha
 
Last edited:
Use HBAs instead of RAID. They are not expensive.
 
Alwin said:
Use HBAs instead of RAID. They are not expensive.
Click to expand...

I already wrote that I do not have such an opportunity. :(
If I can convince the leadership, then it will be paid to the budget for the next year.
And the problem needs to be solved now.

Gosha
 
gosha said:
I already wrote that I do not have such an opportunity. :(
If I can convince the leadership, then it will be paid to the budget for the next year.
And the problem needs to be solved now.

Gosha
Click to expand...
When I was younger I used to work at a car mechanic shop, and I was to use my own tools. I was broke so I only bought the really cheap ones, until one day a wrench broke as I was pushing on it, causing my hand to smash into something and broke my knuckle.

Its a really simple moral- if you go to production with something that you KNOW is faulty, you are going to hurt yourself and your bosses will rightly blame you.
 
  • Like
Reactions: Alwin
alexskysilk said:
When I was younger I used to work at a car mechanic shop, and I was to use my own tools. I was broke so I only bought the really cheap ones, until one day a wrench broke as I was pushing on it, causing my hand to smash into something and broke my knuckle.

Its a really simple moral- if you go to production with something that you KNOW is faulty, you are going to hurt yourself and your bosses will rightly blame you.
Click to expand...

I'm not interested to listen to sermons, but it is very interesting to hear the answer to my last question about possibility of tuning the settings for HDD.

Gosha
 
Last edited:
Hi Gosha,
I am worried because my setup is similar to yours - my disks are on raid cards and configured as 16 single disk raid 0s. That is the closest I could get to un-raided too. I don't get your problem perhaps because I am on proxmox 5.1 ... anyway ...

Have you tried making your osds as filestore? :
pveceph createosd /dev/sdj --bluestore 0
Regarding smart errors, you can sometimes ask the drive for its health through the raid controller :
smartctl -a -d sat+megaraid,3 /dev/sdj
The command is a bit odd, the number (3 here) chooses the disk, the /dev/sdj bit is a space filler.

Thanks for making me aware of the potential problem for my system.
Andrew
 
Last edited:
Andrew Hart said:
Hi Gosha,
I am worried because my setup is similar to yours - my disks are on raid cards and configured as 16 single disk raid 0s. That is the closest I could get to un-raided too. I don't get your problem perhaps because I am on proxmox 5.1 ... anyway ...

Have you tried making your osds as filestore? :
pveceph createosd /dev/sdj --bluestore 0
Regarding smart errors, you can sometimes ask the drive for its health through the raid controller :
smartctl -a -d sat+megaraid,3 /dev/sdj
The command is a bit odd, the number (3 here) chooses the disk, the /dev/sdj bit is a space filler.

Thanks for making me aware of the potential problem for my system.
Andrew
Click to expand...

Hi!
I did not try to make OSDs as filestore.
But I thus used the OSDs in the previous version of ceph (in PVE 4.x).
And all the same HDDs worked without errors.

This don't works for my RAID-controller:

Code: 
smartctl -a -d sat+megaraid,3 /dev/sdb

smartctl 6.6 2016-05-31 r4324 [x86_64-linux-4.15.17-3-pve] (local build)
Copyright (C) 2002-16, Bruce Allen, Christian Franke, www.smartmontools.org

Smartctl open device: /dev/sdb [megaraid_disk_03] [SAT] failed: cannot open /dev/megaraid_sas_ioctl_node or /dev/megadev0

Gosha
 
Hi!

Andrew Hart said:
Have a look in man smartctl there are a few drivers, perhaps yours is hp cciss?
Click to expand...

The result for the disk most often mentioned in error messages:

Code: 
smartctl -a -d sat+cciss,3 /dev/sdb
smartctl 6.6 2016-05-31 r4324 [x86_64-linux-4.15.17-3-pve] (local build)
Copyright (C) 2002-16, Bruce Allen, Christian Franke, www.smartmontools.org

=== START OF INFORMATION SECTION ===
Device Model:     MB2000GFDSH
Serial Number:    9692K0BXFSXC
LU WWN Device Id: 5 000039 74b780f7c
Firmware Version: HPG2
User Capacity:    2,000,398,934,016 bytes [2.00 TB]
Sector Size:      512 bytes logical/physical
Rotation Rate:    7200 rpm
Form Factor:      3.5 inches
Device is:        Not in smartctl database [for details use: -P showall]
ATA Version is:   ACS-3 T13/2161-D revision 5
SATA Version is:  SATA 3.0, 6.0 Gb/s (current: 6.0 Gb/s)
Local Time is:    Wed Jun 27 19:30:33 2018 +05
SMART support is: Available - device has SMART capability.
SMART support is: Enabled

=== START OF READ SMART DATA SECTION ===
SMART Status not supported: Incomplete response, ATA output registers missing
SMART overall-health self-assessment test result: PASSED
Warning: This result is based on an Attribute check.

General SMART Values:
Offline data collection status:  (0x82)    Offline data collection activity
                    was completed without error.
                    Auto Offline Data Collection: Enabled.
Self-test execution status:      (   0)    The previous self-test routine completed
                    without error or no self-test has ever
                    been run.
Total time to complete Offline
data collection:         (  120) seconds.
Offline data collection
capabilities:              (0x7b) SMART execute Offline immediate.
                    Auto Offline data collection on/off support.
                    Suspend Offline collection upon new
                    command.
                    Offline surface scan supported.
                    Self-test supported.
                    Conveyance Self-test supported.
                    Selective Self-test supported.
SMART capabilities:            (0x0003)    Saves SMART data before entering
                    power-saving mode.
                    Supports SMART auto save timer.
Error logging capability:        (0x01)    Error logging supported.
                    General Purpose Logging supported.
Short self-test routine
recommended polling time:      (   2) minutes.
Extended self-test routine
recommended polling time:      ( 222) minutes.
Conveyance self-test routine
recommended polling time:      (   2) minutes.
SCT capabilities:            (0x0025)    SCT Status supported.
                    SCT Data Table supported.

SMART Attributes Data Structure revision number: 16
Vendor Specific SMART Attributes with Thresholds:
ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE      UPDATED  WHEN_FAILED RAW_VALUE
  1 Raw_Read_Error_Rate     0x000f   100   100   050    Pre-fail  Always       -       0
  2 Throughput_Performance  0x0007   100   253   050    Pre-fail  Always       -       0
  3 Spin_Up_Time            0x0003   100   100   002    Pre-fail  Always       -       3221
  5 Reallocated_Sector_Ct   0x0033   100   100   010    Pre-fail  Always       -       0
  7 Seek_Error_Rate         0x000f   100   100   050    Pre-fail  Always       -       0
  8 Seek_Time_Performance   0x0005   100   100   050    Pre-fail  Offline      -       0
  9 Power_On_Hours          0x0032   078   078   000    Old_age   Always       -       8827
 10 Spin_Retry_Count        0x0013   100   100   030    Pre-fail  Always       -       0
180 Unknown_HDD_Attribute   0x003b   100   100   001    Pre-fail  Always       -       0
194 Temperature_Celsius     0x0022   100   100   000    Old_age   Always       -       25 (Min/Max 19/49)
196 Reallocated_Event_Count 0x0033   100   100   010    Pre-fail  Always       -       0

SMART Error Log not supported

SMART Self-test Log not supported

SMART Selective self-test log data structure revision number 1
 SPAN  MIN_LBA  MAX_LBA  CURRENT_TEST_STATUS
    1        0        0  Not_testing
    2        0        0  Not_testing
    3        0        0  Not_testing
    4        0        0  Not_testing
    5        0        0  Not_testing
Selective self-test flags (0x0):
  After scanning selected spans, do NOT read-scan remainder of disk.
If Selective self-test is pending on power-up, resume after 0 minute delay.

Gosha
 
Hi!

It seems to me that I have found a solution to the problem.

ceph.conf - in section [global]:
Code: 
osd scrub sleep = 0.2
osd scrub chunk min = 1
osd scrub chunk max = 2

This increased the number of deep-scrub operations, but reduced the negative impact on the work of all OSDs.

and half cache size in section [osd]:
Code: 
bluestore_cache_size_hdd = 536870912

Now my logfile looks like this:
log.png

Number of deep scrub operations per week:

# ceph pg dump | grep active | awk '{print $23}' | sort | uniq -c
dumped all
104 2018-07-05
129 2018-07-06
258 2018-07-07
268 2018-07-08
73 2018-07-09
192 2018-07-10

Sum = 1024 - as number PGs in storage.
Everything seems to be right :)

Best regards,
Gosha
 
  • Like
Reactions: RobFantini
You must log in or register to reply here.

About

The Proxmox community has been around for many years and offers help and support for Proxmox VE, Proxmox Backup Server, and Proxmox Mail Gateway.
We think our community is one of the best thanks to people like you!

Quick Navigation

Home Get Subscription Wiki Downloads Proxmox Customer Portal About

Get your subscription!

The Proxmox team works very hard to make sure you are running the best software and getting stable updates and security enhancements, as well as quick enterprise support. Tens of thousands of happy customers have a Proxmox subscription. Get yours easily in our online shop.

Buy now!
Community platform by XenForo® © 2010-2024 XenForo Ltd.
Top Bottom
Back