Bug #48412
closedrbd du performance regression
0%
Description
We found Octopus rbd du regression:
Luminous 12.2.13:
[root@ceph-mon0 k0ste]# time rbd du replicated_rbd/volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 NAME PROVISIONED USED volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 1.95TiB 1.76TiB real 0m0.378s user 0m0.349s sys 0m0.008s
Octopus 15.2.6:
[In reply to k0ste] [root@localhost ~]# time rbd du replicated_rbd/volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 NAME PROVISIONED USED volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 2.0 TiB 1.8 TiB real 4m59.940s user 0m13.285s sys 0m10.808s
Vitaliy also checked Nautilus, and it's not affected, so Octopus only regression.
I was collected some logs with --debug_rbd=20
Updated by Vitaliy Filippov over 3 years ago
I have to note that older versions of rbd du are very slow on SOME of the images for me, too. I guess they're slow on images that are almost 30 TB large. I ended up writing my own node.js utility to get used space... it just downloads rbd_object_map's using RADOS and it's always fast.
Updated by Konstantin Shalygin over 3 years ago
- Target version deleted (
v15.2.7)
Updated by Konstantin Shalygin over 3 years ago
Also 15.2.7 debug logs:
https://drive.google.com/file/d/19GpT07JLLJGfoIiibwt6b6S3CXglsiqr/view?usp=sharing
Updated by Konstantin Shalygin over 3 years ago
I was rerun's `rbd du`, and gets significantly different results, don't know the real reasons what it depends on.
[root@localhost ~]# time rbd --debug_rbd=20 du replicated_rbd/volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 >> /root/rbd_15.2.7_debug 2>&1 real 1m28.532s user 0m13.723s sys 0m11.656s [root@localhost ~]# time rbd du replicated_rbd/volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 NAME PROVISIONED USED volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 2.0 TiB 1.8 TiB real 5m19.817s user 0m13.343s sys 0m11.194s [root@localhost ~]# time rbd du replicated_rbd/volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 NAME PROVISIONED USED volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 2.0 TiB 1.8 TiB real 6m55.623s user 0m12.352s sys 0m10.414s [root@localhost ~]# time rbd du replicated_rbd/volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 NAME PROVISIONED USED volume-e5267e8a-3a18-4c27-842e-680f1df1beb3 2.0 TiB 1.8 TiB real 6m24.897s user 0m12.226s sys 0m10.515s
Updated by Mykola Golub over 3 years ago
- Subject changed from octopus: rbd du performance regression to rbd du performance regression
- Status changed from New to Fix Under Review
- Assignee set to Mykola Golub
- Backport set to octopus
- Pull request ID set to 38389
Updated by Jason Dillaman over 3 years ago
- Status changed from Fix Under Review to Pending Backport
Updated by Mykola Golub over 3 years ago
- Copied to Backport #48470: octopus: rbd du performance regression added
Updated by Nathan Cutler over 3 years ago
- Status changed from Pending Backport to Resolved
While running with --resolve-parent, the script "backport-create-issue" noticed that all backports of this issue are in status "Resolved" or "Rejected".