Absolutely Horrible Response Times

[stanj]

Originally Posted by Brangwyn

I concur, I've just moved a website from one server to another, no difference, moved the database off mySQL8 and bingo the site is back to normal. someone needs to take a close look at these mySQL servers, heck with this many in production you'd think an Enterprise MySQL Subscription wouldn't be too much of a stretch so they could use the great analyzer, alerting and monitoring tool the MySQL has.

My php/MySql driven sites on NT45 have been painfully slow and timing out during some part of each day but the static sites(we only run a couple sites on a reseller account for just one business, not for reselling) seem to be ok(except the recent rash of Application Pool crashes that cause Service Unavailable almost daily). This slow-down has been going on for a long time however, at least 1.5 years. I think you are right about much of the slowness is in the MySql realm. I finally ordered a VPS with resources far in excess of my needs just to make sure the performance problems of our shopping cart would be resolved. The full switch has not been made but I am monitoring and testing the new server. I notice the trace route shows the new account has a few more hops but overall the performance of both systems are quite good through the switches, even static page display(NT45) is not much different but the shopping cart doing a report or other intensive DB process, there is a dramatic difference, 100 times faster have been measured doing tests during low usage times(3am EST) and the tests can not reliably be done during peak US times because of time-outs on Mysql5 queries.
No doubt in my mind, there is problem in the system. Maybe there is a natural limit to how large a h-sphere cluster can be and still maintain liability. Would splitting the cluster into smaller self contained units help? There would be higher licensing fees for more small clusters but if that helped make things more reliable I doubt anyone would complain if it cost a bit more. No one is complaining of prices, if anything they might be too low as they are for maintaining the system.
Just my thoughts as a comparative amateur.

[dvanburen]

Originally Posted by antic

Thanks for the post-mortem. But if this had been explained at the beginning, there wouldn't have been a problem, would there? That's the difference between a service and "service".

It's assumed that index pages are set properly. Server moves are not routine and are certainly not recommended for reasons which you yourself have experienced. That said, I will see if we can revise the pre-defined reply yo include this warning, though it is of little use unless it is set before the transfer is started.

[antic]

Originally Posted by dvanburen

It's assumed that index pages are set properly. Server moves are not routine and are certainly not recommended for reasons which you yourself have experienced. That said, I will see if we can revise the pre-defined reply yo include this warning, though it is of little use unless it is set before the transfer is started.

Thanks. Yes the CMS on that site was a bit unusual, so the index.html that HSphere created overrode the usual site behaviour. But if there was a support ticket response just to say "it's done please check it", that would help a lot. Perhaps with a link to a "site move info" page, which points out that site files are copied over the top of a normal HSPhere setup, which means files like index.html, login.html and the CP & CGI-BIN folders will reappear. That's important, as login.html, CP & CGI-BIN folders should be deleted on most production sites to avoid hacking/fiddling. If it's an ASP or PHP site, then index.html should be removed too. It's just a "procedures" thing that would help, but then I'm obsessive about stuff like that.

Anyway I didn't mean to hijack this thread, other people here have more pressing concerns to respond to.

[tkraffty]

One of my clients on the CF machine reported a "slow down" starting this morning, and now their site has been down for over 4 hours. Other sites are a bit slow too, but still operational... Email issues up the wazoo...

Phone calls to V are unanswered. Emails aren't even being received, let alone answered -- most likely due to some huge volume of spam which appears to be OUTGOING from V's mail servers (see my last post).

There's obviously a whole lotta holy crap going on there... Perhaps some server updates went bad, or there's a router problem (doesn't explain email spam issues).

Either way, as a long-time Vortech reseller, who's investing many thousands in Vortech over the years, I'd appreciate an email update or forum post describing what is happening across your network and data-center. There is obviously SOMETHING big going on. Might as well speak up. Your new customers are going to find out about the ongoing company issues sooner or later anyway... And the rest of us "old timers" are used to the poor communication (although never happy about it).

That's all.
TK