Last nerve.

This is a rant. I'm not asking for advice, unless somebody can recommend some good shared hosting companies for me to check out.

My server has stabilized to some degree since the summer fiasco was mostly resolved. I don't often notice load spikes to 25 and above now, though they still occasionally happen. My sites are up... most of the time. Maybe 1-2 times a week they're down. (For those playing along at home, that's still not acceptable.)

However, I'm now getting unexplained Internal Server Errors which come and go -- I suspect a flaky database connection -- as well as random bouts of painful load times. There is also some kind of problem with e-mail at least once a day. Most often the IMAP server temporarily "forgets" my passwords, meaning I have to keep typing them until it recovers. Or it loses track of what's in the mailbox for a bit and then later on it might take forever to redownload headers. Or SMTP relaying fails, and when it works it's ALWAYS slow. "Use webmail" is not a solution.

A few days back I contacted support about ongoing issues with my server and they offered to move me, which I declined for the moment because I know such a move would inevitably break things which I don't have time to fix at the moment. They said they'd put my server on their administrative to-do list which means monitoring it and eventually giving it a round of upgrades as needed. Well, that appears to have happened because I spent half the night tonight getting my 'screen' processes killed while working on one of my sites. Later on, a 'make' process which support RECOMMENDED I DO got killed as well -- I was trying to build a custom PHP5 binary so I could use outbound sockets. Fortunately I just commented out half the build script and was able to start over and it didn't get killed again.

I rely on my sites to do homework, and I have a group of about 30 college students which also rely on my sites to do homework. Random outages are not acceptable. Oddly enough, though, it's the e-mail problems that piss me off the most because they're so frequent and they've been going on for months.

If I didn't have so much crap deeply intertwined with Dreamhost I'd likely already have bailed. The only thing that's keeping me here is the large amount of pain I'll have to go through to move somewhere else, and the fact that Josh or whoever was spot on in the Dreamhost blog article about crappy "web hosting review" sites. There are literally hundreds of such sites which are complete BS and I've been hard pressed to find a real one with information that isn't totally suspect.

I'm looking, though.

In reply to:

---

My server has stabilized to some degree since the summer fiasco was mostly resolved. I don't often notice load spikes to 25 and above now, though they still occasionally happen. My sites are up... most of the time. Maybe 1-2 times a week they're down. (For those playing along at home, that's still not acceptable.)

---

Your web machine's load is higher then it should be, but it has not been rebooted for 2 weeks now:
digdug: 12:39 PM# uptime
12:39:24 up 14 days, 22:12, 4 users, load average: 7.36, 7.36, 6.59

I have assigned someone to move a few of the heaviest users off of your server today. Also you should have accepted the offer to move you to a different server. Our system for moving people between servers is used all the time and works very well. Unless you have a non-standard setup such as having your dns service set up with a different company it is fully automated.

Can you please make a list of the exact email problems you have had. That way I can match them up with any known issues that have been resolved and make sure there are no outstanding ones that need attention today.

All my mail, as far as I know, is hosted on murdock. Some of my boxes are mail-only and some are related to shell accounts. Sometime late Friday early Saturday the mail server "forgot" my passwords as it tends to do fairly often. In Apple Mail and Thunderbird, I was being asked for my password over and over. After a while, all the passwords "took" except for m6992695 which took until after Saturday morning sometime to recover.

The "forgetting" of passwords on one or more mailboxes, sometimes just once and sometimes for long periods, is an almost weekly occurance which usually appears to resolve spontaneously, sometimes after periods longer than a "hiccup" of a minute or so. Of course every time there's a "hiccup" I have to type all my passwords back in by hand.

Other issues which I experience commonly but weren't happening within the past 2 days to my knowledge are errors sending via SMTP -- server drops connection or seems to die when a message is sending, or sending messages with attachments is glacially slow compared to how fast it should be over my broadband. Less common but definitely not unheard of are slow attachment downloads via IMAP and slow downloads of headers via IMAP.

I use the old-messages copier to keep my inboxes fairly trimmed.

Thanks for the attention in the forums. I've had 2 tickets open now, one for 27 hours and one for almost 22, regarding the acute issues. They're 1545726 and 1545542.

As for the server move, I have a couple off-site daily backup scripts which I didn't have time to update when I was first offered the move, and I tend to be paranoid about such things even when told it's simple, etc. I currently have all my sites on digdug, and if moving some power users off of it helps things out I'll be content to stay there. I was also told I'd have a couple hours of downtime with a move which I wanted to avoid, or if it was needed, schedule during the wee hours of Eastern time if that's possible.

Did you try it using a different email client? I'm on the same server for email and using Eudora 7.0.1.0. In a year on that server I've never had a single problem, with the exception of maybe 5 or 10 minutes where the server's inaccessible (siteuptime.com agrees with me), but... eh, it's only 10 minutes.

Yes, I have an opinion.

Get a minimum 50% off with the "haveadreamyday" promo code, and... have a dreamy day. Original, no?

The same behavior occurs on 2 different clients on 2 different systems and I've seen it from more than one location.