Rack Admins

GFS Cluster is good thing. It simple and robust: If first node down, secondary replace it.

But… Did you know about GFS work? It write all data beetween servers via network connection. Usually servers connect via 1Gbps – it is 128Mb per second. Sounds good? Please hold on! This is a RAW bandwith. TCP/IP connection take 5%. GFS take 10%.

Well, 100mb per second sound good, but one second! GFS is a beedirectional protocol such like DRBD. So, it split channel twiced – 50mb per second.

And now check – did you MySQL will work nice for that drive speed? Of course Not!

But we have uniqe techniqe for solution this problem: Use secondary node as dedicated network block device. It avoid spliting channel and you got full 100mb per second.

Sounds good, yea?

[si-contact-form form='2']

Popularity: 3% [?]

Sometime you may got strange problem. Server partition, usually /var or /tmp went to read only.

Some peoples suggest you unmount that partition and use main (root) partiton for place this directory. This is very big mistake!

Partition become to read only if you got some fiilesystem errors. It may be software or hardware problem.

and since the partition is mounted on the main disk it needs to be replaced, so i need someone to perform the following work:

The cure is not to easy:

1. Detect what same cause this error – software or hardware
2. Software problems cure tune2fs and kernel upgrade
3. Hardware problems cure drive replacement

But really disaster problem – error on drive contorl or main board. So, drive replacment not help you. We suggest migrate server to new one. We can do it for you. Zero downtime guranteed:

Backup of the all accounts.
Once the datacenter sent access to new server we configure WHM/Cpanel or other Hosting Panel.
Restore of the all accounts to new server.
Check Apache & MySQL conditions.
Redirect all traffick from old to new one.
Cange DNS settings.

We use our know-how for traffick redirection. It absolutly transparent!
[si-contact-form form='2']

Popularity: unranked [?]