server migration disaster

[darcy]

About a week ago I noticed two of 5 VPS servers (3 at DEHE, 2 at JagPC) were noticeably slower than they had been. I did some diagnosis and noticed that I was getting little CPU time and very high iowait % (pegged at 100% for minutes at a time) on these two servers where previously I had not. Both servers were on the UL3 node at DEHE. I complained and after a bit DEHE offered to move both servers to a new node (UL10). Nice ! Or so I thought.

Unfortunately one the servers did not migrate as expected. 3 days later I am still without my mail server. The story is that Plesk got hosed during the migration. After the fact I find out the weekly backups of the node I thought were happening were not actually happening. I was doing my own daily backups but not of the Plesk config unfortunately.

The ticket on the issue is a mile long. Getting any real information, apologies, assistance or understanding from DEHE has been like getting blood from a stone. I've been told that the migration problem has been raised with SWSoft and that I should wait - wait for what ? My server has been hosed for 3 days - I frankly care little about what actually went wrong. Forget SWSoft and help me get my mail server back up for cripes sake!

The bottom line is as I type this, my mail server is trashed as a result of the botched migration and DEHE has yet to offer me any sort of compensation (they have hinted at but not defined "compensation"). I'm trying to get them to set me up a free server with Plesk so that I can try to recreate my email accounts, but I have yet to get them to commit to this. You'd think that would be a reasonable way to convince me that I should stay with DEHE, right ? We'll see what happens.

[Robertjp]

Hi,

Yes we understand that migration for your vps that was done by us didn't went successful and caused problems but we tried our best to get this fixed ourselves and through the Virtouzzo's Vendor Swsoft. We still have a ticket opened with them to check on the reason that why the VPS got removed from the source node, which caused the real problem as otherwise we may simply have started your vps on the source node UL3. As we moved the VPS with the option to keep it on the old node as well , hopefully we will receive a response from Swsoft soon to see what went wrong.

As I have already updated you in your ticket that all your user data is safe and as soon as you confirm we are going to backup your data and setup a new VPS. I look forward for your confirmation in your support ticket.

Further referring to your feedback regarding compensation, we will definitely compensate you for these problems and the downtime you have faced.

This is one of the rare case where the migration didn't went successful , please accept my apologies for the same.

[darcy]

I accept your apologies. I fully understand that this was an unforeseen problem and at a technical level is SWSoft's responsibility. My point is that I don't really care about why it happened. I just want someone at your end to take responsibility for getting my mail server back to operational status. I should note that as of today (4th day of downtime) this is finally happening, but it's has taken way too my effort on my part to stir up any action.

[darcy]

If your main business email went down, how many days do you think it would take you to get it back up and running? I don't think it would be 4 days.

[Robertjp]

Your vps on old node was logically deleted but we were able to track its physical copy and your old VPS was brought live yesterday on the old node UL3. Our second migration try also faced some problems and we are discussing all the aspects in detail with Swsoft to make sure there are no such issues in any future migrations. Once we have detail info from them and we test that , we will migrate your vps to the new node and till then it will keep running from the old Node. For any issues you may have please let us know through your support ticket.

[darcy]

ok - that's good - today I've finally seen the sort of action on this issue that I expected 4 days ago.

[darcy]

I like how this thread was moved from the testimonials section to this dark neck of the woods...priceless. Anyway, here's an update - server was restored on the old node eventually. Turned out to be a file corruption issue was breaking the migration process somehow. My advice is to make sure that DEHE do a backup just before they migrate your VPS.

Of course as I type this I seem to be going through the same sort of routine with another migration a week later. The gods of virtuozzo do not smile on me apparently. It's like deja vu actually.

[darcy]

All is good now -- in summary, all my servers have been moved to a new node and they are noticeably faster. My observation is that over time, presumably as DEHE adds customers to nodes, server performance goes down. It appears to up to the customer to monitor server performance and then complain if things get too overloaded. In my case high IOWAIT% has been the killer each time -- moving to a new node has brought IOWAIT down to about 0%. To their credit DEHE seem to be willing to migrate servers over to new nodes if things get too problematic.

[JPC-Greg]

I like how this thread was moved from the testimonials section to this dark neck of the woods...priceless. Anyway, here's an update - server was restored on the old node eventually. Turned out to be a file corruption issue was breaking the migration process somehow. My advice is to make sure that DEHE do a backup just before they migrate your VPS.

Of course as I type this I seem to be going through the same sort of routine with another migration a week later. The gods of virtuozzo do not smile on me apparently. It's like deja vu actually.

I moved it because it wasnt a testimonial, more of a support request that needing some addressing. People dont really reply to testimonials.

[JPC-Greg]

All is good now -- in summary, all my servers have been moved to a new node and they are noticeably faster. My observation is that over time, presumably as DEHE adds customers to nodes, server performance goes down. It appears to up to the customer to monitor server performance and then complain if things get too overloaded. In my case high IOWAIT% has been the killer each time -- moving to a new node has brought IOWAIT down to about 0%. To their credit DEHE seem to be willing to migrate servers over to new nodes if things get too problematic.

We dont leave it to clients to report it but it helps if you do , especially if its something we've missed. As I prepare the dehe merge with jag we setup extra servers we had in jag to help the dehe servers that had the high i/o issues. It helps to be able to steal machines from big brother . Anyhow we are putting out some new software and policies that will vastly improve and limit the new account i/o delima. It will soon become a null point.

Thanks for your patience through that issue though, we greatly appreciate your business.