OVH OVH Public Cloud Servers ?

Finally got to test their "high performance" storage myself. Pretty much the same results as you @eva2000

Code:

[root@host ~]# mount | grep /dev/vdb1
/dev/vdb1 on /mnt/db type xfs (rw,relatime,attr2,inode64,noquota)
[root@host ~]# df -h
Filesystem      Size  Used Avail Use% Mounted on
/dev/vda1       197G   33G  157G  18% /
devtmpfs        3.4G     0  3.4G   0% /dev
tmpfs           3.4G     0  3.4G   0% /dev/shm
tmpfs           3.4G  281M  3.1G   9% /run
tmpfs           3.4G     0  3.4G   0% /sys/fs/cgroup
tmpfs           682M     0  682M   0% /run/user/1000
tmpfs           1.0G     0  1.0G   0% /var/mysqltmp
tmpfs           682M     0  682M   0% /run/user/0
/dev/vdb1        10G   33M   10G   1% /mnt/db
[root@host ~]# cd /mnt/db/
[root@host db]# ls
[root@host db]# ioping -c 10 .
4 KiB from . (xfs /dev/vdb1): request=1 time=352 us
4 KiB from . (xfs /dev/vdb1): request=2 time=868 us
4 KiB from . (xfs /dev/vdb1): request=3 time=319 us
4 KiB from . (xfs /dev/vdb1): request=4 time=315 us
4 KiB from . (xfs /dev/vdb1): request=5 time=297 us
4 KiB from . (xfs /dev/vdb1): request=6 time=316 us
4 KiB from . (xfs /dev/vdb1): request=7 time=254 us
4 KiB from . (xfs /dev/vdb1): request=8 time=312 us
4 KiB from . (xfs /dev/vdb1): request=9 time=276 us
4 KiB from . (xfs /dev/vdb1): request=10 time=360 us

--- . (xfs /dev/vdb1) ioping statistics ---
10 requests completed in 9.01 s, 2.73 k iops, 10.6 MiB/s
min/avg/max/mdev = 254 us / 366 us / 868 us / 169 us
[root@host db]# ioping -RD . 

--- . (xfs /dev/vdb1) ioping statistics ---
3.31 k requests completed in 3.00 s, 1.11 k iops, 4.36 MiB/s
min/avg/max/mdev = 78 us / 896 us / 116.0 ms / 2.90 ms
[root@host db]# ioping -RL .

--- . (xfs /dev/vdb1) ioping statistics ---
1.48 k requests completed in 3.00 s, 511 iops, 127.9 MiB/s
min/avg/max/mdev = 1.46 ms / 1.95 ms / 11.3 ms / 408 us
[root@host db]# ioping -RC .

--- . (xfs /dev/vdb1) ioping statistics ---
2.88 M requests completed in 3.00 s, 1.14 M iops, 4.35 GiB/s
min/avg/max/mdev = 0 us / 0 us / 37.8 ms / 23 us
[root@host db]# dd if=/dev/zero of=sb-io-test bs=1M count=1k conv=fdatasync
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 1.94348 s, 552 MB/s
[root@host db]# dd if=/dev/zero of=sb-io-test bs=64k count=16k conv=fdatasync
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 1.75944 s, 610 MB/s
[root@host db]# dd if=/dev/zero of=sb-io-test bs=1M count=1k oflag=dsync
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 15.1472 s, 70.9 MB/s
[root@host db]# dd if=/dev/zero of=sb-io-test bs=64k count=16k oflag=dsync
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 93.8036 s, 11.4 MB/s
[root@host db]# 

---

 

Read this on a site:

So:

Code:

mount -o nobarrier /dev/vdb1 /mnt/db

Code:

[root@host /]# mount | grep vdb1
/dev/vdb1 on /mnt/db type xfs (rw,relatime,attr2,nobarrier,inode64,noquota)
[root@host /]#

Code:

[root@host db]# dd if=/dev/zero of=sb-io-test bs=64k count=16k oflag=dsync
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 10.6481 s, 101 MB/s
[root@host db]# dd if=/dev/zero of=sb-io-test bs=1M count=1k conv=fdatasync
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 1.68721 s, 636 MB/s
[root@host db]# dd if=/dev/zero of=sb-io-test bs=64k count=16k conv=fdatasync
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 1.82256 s, 589 MB/s
[root@host db]# dd if=/dev/zero of=sb-io-test bs=1M count=1k oflag=dsync
1024+0 records in
1024+0 records out
1073741824 bytes (1.1 GB) copied, 3.03999 s, 353 MB/s
[root@host db]# dd if=/dev/zero of=sb-io-test bs=64k count=16k oflag=dsync
16384+0 records in
16384+0 records out
1073741824 bytes (1.1 GB) copied, 10.6453 s, 101 MB/s
[root@host db]# 

 

"should" be OK with the nobarrier option enabled, given the triple write setup they have for their storage.

 

Good question. I've set it up with nobarrier, and moved MySQL over to it (purely on my own server running my site). From what I can find, it would only cause corruption on a single drive failure.

Even percona sets nobarrier on their guide to XFS
Setting up XFS on Hardware RAID - the simple edition - MySQL Performance Blog

On a side note, would it be possible to have an option set in centmin.sh for the MySQL socket location? Currently it's hard set.

 

I changed the my.cnf config to use the new socket location I set on the new drive, but websites were failing to load with "an unexpected error occured", and mysql command line didn't work either

Code:

[root@host databases]# mysql -u root
ERROR 2002 (HY000): Can't connect to local MySQL server through socket '/var/lib/mysql/mysql.sock' (2 "No such file or directory")
[root@host databases]#

 

Yeah, literally just figured [client] wasn't set anywhere

 

Code:

sed -i 's|/var/lib/mysql/|/mnt/db/mysql/|' php_configure.inc

and rebuild PHP has fixed it.

 

Takes 38 minutes for the cPanel backup to complete, and send the files off server when it's done.

 

The IO on my Cloud VPS is in the toilet for the last 2 weeks. I've got a ticket open with them currently.

 

They never fixed it, so the cloud VPS is gone. Moved over to Ramnode now.