Discussion for New Server Platform & Features Announcement

[H2O]

I'm getting a register globals error on oscommerce.. howcome???

[AndrewT]

register_globals is now disabled by default on our new servers. You will have to create your own php.ini file in the directory of the script with the setting enabled if required.

[-Oz-]

before you ask: make a file called php.ini and put this line in it:

register_globals = On

and upload it to your public_html folder.

[AndrewT]

The php.ini file must be placed in the actual directory of the script that requires the setting. This does not recursively go through subdirectories.

[Frank Hagan]

One other issue ... I had one account that had "Front Page Extensions" added to it, and for some reason the Front Page widgets (or whatever that Satan's Spawn at Microsoft called them) quit working ... things like the "hit counter", etc. Adding the Front Page Extensions to the domain through WHM solved the problem.

[smartcard]

How long the migration normaly take from the old server to new?

[AndrewT]

Complete data transfers have taken anywhere from a 1 day to nearly 3 days. I would guess that cpanel04 will be on the longer side of that given the hardware induced performance problems.

[SitesByRob]

Hey Andrew,

Assuming you can recover all data off the old cpanel04 drives, if we restored a user's site ourselves will you overwrite with the recovered structure? For me that's fine - I just want to know to expect it.

Also, I noticed some issues getting cgiemail to work last night on new cpanel52. Not sure if it is at all related.

[AndrewT]

We're only going to restore user data upon request since many customers have already uploaded and/or updated their sites with the appropriate content we certainly do not want to overwrite these changes and create further problems.

In regards to cgiemail, I'm not entirely sure as to what you are referring to but if you are still experiencing problems please submit a ticket.

[SitesByRob]

Thanks!

As an anxious user of cpanel04 I want to thank Andrew and his team for working hard to save our files and restore our sites! Thank you!!!!

[scojam]

must we request data restore?

Do we need to request data restore for our WHM and Cpanel accounts?

I would like to help the image transfered directly, instead of me having to upload the sites again. But does this image transfer require a work order request?

thanks,
Scott

[AndrewT]

This was only an issue with cpanel04 due to hardware failure. All other servers that will be or have been upgraded include complete data transfer automatically done by us.

[CasualObserver]

Cpanel16 server move cancelled after it was almost completed...

I just saw the update where the CPanel16 server move was canceled after it was basically completed due to the wrong drives being installed in the replacement server.

I just went through 5 to 6 hours of not having email access for myself or client domains since the server was not able to handle the stress of copying the data off and also doing email. Andrews reply was that it would be sporadic. Believe me, was not sporadic. It was completely inaccessible email both for sending and receiving through Dathorn. I had to deal with my clients for most of the day concerning this and I do not want to repeat this.

It is generally considered bad business practices to perform server moves during peak business hours that can be avoided. Since these are scheduled moves that seem to happen once a week, I simply can not understand how these can not be scheduled to start during the evening when usage should be minimized and possible downtime for clients also minimized. Or even better during a Friday night / Saturday time window.

It seems that this move should have taken between 6 to 9 hours to complete. Given that time frame, if this move was started during the late evening hours, it should be completed before the start of business for local customers. Say midnight EST with a completion window of about 9AM EST to possibly noon EST.

I really do not want to have to explain to people why they will be down for another day next week if the second attempt does the same thing and results in lack of email during what amounts to almost another complete business day.

Saying sporadic email problems is not the same as not being able to have a server respond once during the entire period of time. That is precisely what happened today.

Please let me know if you plan on possibly handling this second attempt at the server move at a more appropriate time for business customers.

Thanks in advance.

[AndrewT]

I do certainly apologize for the inconvenience.

Most of the server upgrades that we've done, aside from a few, take 24-36 hours to complete. Some of them even take a bit longer so it is not quite as simple as just doing it in the middle of the night. We would certainly do this if it could be completed in one night.

You must also realize that we provide hosting for far more than just U.S. customers. A good 40+% of our customers do not even reside in North America. No matter what time we schedule an upgrade it will always impact someone's "business hours".

Ultimately it is the older SATA or IDE drives in some of our servers that are killing overall performance, especially when we are attempting to transfer all accounts on the server in addition to normal activity. Under very high loads like this, Exim is generally the first service to show problem signs with POP and IMAP following shortly thereafter. Our goal is to complete these upgrades as quickly and painlessly as possible. Our newer servers perform significantly better than the older ones and these performance problems won't be present on the new platform.

[CasualObserver]

I do certainly apologize for the inconvenience.

Most of the server upgrades that we've done, aside from a few, take 24-36 hours to complete. Some of them even take a bit longer so it is not quite as simple as just doing it in the middle of the night. We would certainly do this if it could be completed in one night.

Since your average rollout seems to be roughly every week or two, this sounds like a perfect situation for starting on a Friday evening and finishing sometime during Saturday or Sunday. In general, I believe business activities tend to be less during evening hours and weekends then during the Monday - Friday weekdays. 24 to 36 hours would finish it easily during a relatively dead period of time.

You must also realize that we provide hosting for far more than just U.S. customers. A good 40+% of our customers do not even reside in North America. No matter what time we schedule an upgrade it will always impact someone's "business hours".

While I did not know the percentage was this high for customers outside of the USA, it still does not change my point. I believe Dathorn operates on either EST or CST time zones for your office. Customers signing up for service know that Dathorn is in the USA. It is standard procedure to do planned maintenance during the period of time where the majority of your customers are least busy. This applies equally to large firms such as eBay, Amazon etc. They always schedule any planned downtime for the evening / middle of the night period for where they are located. Both also deal a large amount in international customers. If I had signed up for a web host provider that was located in Europe I would fully expect them to do maintenance and planned upgrades during their middle of the night periods which would be mid morning for me in EST time zone. If it was an Australian provider, I expect them to be doing it during the middle of my day.

Ultimately it is the older SATA or IDE drives in some of our servers that are killing overall performance, especially when we are attempting to transfer all accounts on the server in addition to normal activity. Under very high loads like this, Exim is generally the first service to show problem signs with POP and IMAP following shortly thereafter. Our goal is to complete these upgrades as quickly and painlessly as possible. Our newer servers perform significantly better than the older ones and these performance problems won't be present on the new platform.

I think we all appreciate the upgrade you are doing. It is going to improve service vastly just from looking at the hardware specs. I actually thank you in advance for it.

That being said, it does not change the timing that you are using in the slightest to complete these scheduled upgrades. What you said reinforces to me that you are aware the moves will negatively impact email performance possibly to the point of being unable to use email during the data migration.

I do mostly LAN and WAN work for local clients that I support onsite. During the last 15 years I have done numerous server moves / upgrades for companies that would either take them completely down or at least severely impact their workflow. I arrange the planned changes to the hours I mentioned above to minimize this impact to the client. It shows a level of respect for your clients to do that. It shows you respect their business and their own support issues that downtime results in. From personal experience I know that doing server migrations during off hours stinks for the people doing it. I also know that makes the clients much more happy and even more dedicated.

I feel you run a very smooth operation here and the service and support has always been excellent. I know the vast majority of people on the boards feel the same way. What I am requesting is simply the "polite and conscientious" way of doing server migrations. It disrupts the least amount of people and reinforces that you care about your customers possible support headaches also when humanly possible.

Thanks again.