General

Profile

Vitaliy Filippov

Issues

Activity

06/15/2019

09:36 AM Ceph Revision 47cd1afa (ceph): osd/bluestore: Actually wait until completion in write_sync
This function is only used by RocksDB WAL writing so it must sync data.
This fixes #18338 and thus allows to actuall...

05/20/2019

04:10 PM bluestore Feature #38816: Deferred writes do not work for random writes
Sage Weil wrote:
> Vitaliy Filippov wrote:
> > 1) Bluestore doesn't honor the max_deferred_txc parameter. It starts...

04/11/2019

08:52 AM bluestore Feature #38816: Deferred writes do not work for random writes
I think we are beginning to discuss something different from the original question, but ...
I checked the code and...

03/25/2019

01:55 PM bluestore Feature #38816: Deferred writes do not work for random writes
I tried similar thing when Mark asked me. In summary, you *can* enlarge your deferred queue a bit, but you can't make...

03/12/2019

03:39 PM bluestore Bug #38559: 50-100% iops lost due to bluefs_preextend_wal_files = false
Yes, I've thought of that but I haven't tested it... However this is rather strange then. Who does the fsync if BlueF...
12:02 PM Ceph Revision c703cf9a (ceph): osd/bluestore: Actually wait until completion in write_sync
This function is only used by RocksDB WAL writing so it must sync data.
This fixes #18338 and thus allows to actuall...

03/06/2019

09:55 PM bluestore Bug #38489: bluestore_prefer_deferred_size_hdd units are not clear
So that's why write_big operations may be also deferred just like write_small's. OK, thank you very much, it's clear now
04:08 PM bluestore Bug #38489: bluestore_prefer_deferred_size_hdd units are not clear
Forgot to mention, this was Ceph 14.1.0
09:25 AM bluestore Bug #38489: bluestore_prefer_deferred_size_hdd units are not clear
I've just tried to set
[osd]
bluestore_prefer_deferred_size_hdd = 4194304
On a test HDD plugged in my laptop. ...

03/03/2019

08:07 PM bluestore Bug #38559 (Resolved): 50-100% iops lost due to bluefs_preextend_wal_files = false
Hi.
I was investigating why RocksDB performance is so bad considering random 4K iops. I was looking at strace and ...

Also available in: Atom