New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
LOL...well, I am an unabashed BuyVM shill (see my site) but I'm not shilling for BuyShared...yet...
To be clear, there've been no hot tubs, no hookers, no cocaine, and no midgets, so unless @Francisco ponies up (har!) he's not getting my endorsement.
Listen, UPS was late this month with the pickup, quit busting my balls.
Francisco
I don't -- I cancelled maybe six months ago. All I can tell you now is that it was in Vegas and with both Rackspace and Pingdom monitoring and my own direct visits there was a lot of downtime and a lot of very slow (like 20 seconds+) loadtimes even for plain HTML pages.
I own 2 buyshared accounts. One in LU2 and one in LV3. LU2 is new and stable imho. LV3 has very minor hickups. I believe Francisco explained in another thread a while ago. Anyways, here I post my uptime report for Vegas03 (LV3). Downtimes are little and usually around 1min. This is a 1min interval check.
Let me know if that smooths things out. The node was running quite low on entropy so I shoved haveged on there to help. With the increase in letsencrypt use it's going to chew through it pretty good.
I also reinstated fail2ban incase it was any large brutes chewing up too much CPU. When cloudlinux had some crash bugs in their kernel that was taking this node offline for 3 - 5 minutes a day in January we turned off fail2ban thinking it was maybe triggering it but we never added it back.
CL7's been stable since the end of January and our other CL7 nodes have been running fail2ban w/o too much issue.
By all means, if you see things like this you're welcome to ticket. Our own monitoring doesn't always catch it since we don't HTTP monitor every single IP we have assigned (we'd be DDOS'ing ourselves every minute).
Thanks
Francisco
Growing pains with shared hosting are definitely a real thing. If I trust anyone to work them out its Fran.
Could it be network hiccup?
Nope, I went through nullroute logs and there was nothing hitting that node. As others have said they have service on different LV services without problem, it's just litespeed supposedly crapping out.
It could be entropy related, haveged is working hard, as is fail2ban. I increased the sensitivity on my own monitoring to see if I see anything else happening. So far its been quiet with plenty of resources spare.
Francisco
Here is what I have for my Las Vegas shared account. The ping seems to be a much lower amount, where as the HTTP checks are a bit higher. It usually doesn't last for very long though, so it's never been a concern for me.
they will denied that
I've asking for few times and they always said it from my side 
I've gotten 2 NodePing alerts on my BuyShared service in the past hour so I've changed the Sensitivity to Low (from Medium) and set the Timeout to 20 seconds (from 5 seconds). Fingers crossed!
Fingers crossed!
[2016-04-25 19:17:29 -0700] info [safeapacherestart] Restart elapsed seconds: 76
[2016-04-25 19:19:07 -0700] info [safeapacherestart] Restart elapsed seconds: 4
[2016-04-25 19:19:58 -0700] info [safeapacherestart] Restart elapsed seconds: 3
[2016-04-25 19:21:36 -0700] info [safeapacherestart] Restart elapsed seconds: 4
[2016-04-25 19:24:20 -0700] info [safeapacherestart] Restart elapsed seconds: 3
[2016-04-25 19:36:25 -0700] info [safeapacherestart] Restart elapsed seconds: 4
[2016-04-25 22:07:12 -0700] info [safeapacherestart] Restart elapsed seconds: 11
[2016-04-25 22:08:50 -0700] info [safeapacherestart] Restart elapsed seconds: 70
The ~ < 10 second ones are fine since that's a rehash and such but the 70 second ones actually tripped my alerts (now that they're very sensitive).
Now to figure out what's making a rehash take that long. I was thinking it was maybe the LE module but it hasn't done any renewals today it looks like.
May just ask the cPanel developers to take a look.
Francisco
@Francisco what time zone is that? My alerts occurred at 11:57PM and 1:11AM EST.
I had email alerts for 8:52 PM & 10:10 PM PDT.
I'm using 1 minute checks for what it's worth.
Francisco
Setup something that checks it every 10 secs, so you are able to detect mostly all small network issues.
It isn't network related since the 2nd alert I had top running and saw litespeed hard reset (all lsphp processes went with it) and it took a few to come back. It restarted perfectly fine, it just took longer than it should've for those 2 times which is odd since the other rehashes took ~4 seconds.
Francisco
I know, but for whatsoever reason i wrote Network issues.
Don't ask why.
I'm really surprised that BuyShared isn't offline with all this info out in the wild... I mean it wouldn't be hard to keep adding and removing domains essentially Dosing them from the inside..
Right? Or is litespeed a bit smarter than that?
This is the only node doing this, we have 7 nodes total at this point.
Litespeed/apache do graceful reloads for its configuration whenever:
Graceful restarts take a couple seconds to apply and at least with litespeed it usually happens seamless since it spawns off a new main process to carry on.
Now we had an issue many months ago where we had 2 - 3 reseller accounts (from the same owner) on a single node that created something like ~40,000 databases and accompanying domains which fattened our InnoDB to wanting ~160GB RAM to fit everything in memory as well as our httpd.conf to ~90MB since there was SSL's involved too. The user was given a backup and asked to leave since it was just some really big amazon affiliate click farm they were running.
Once we had that happen though I wrote some audit scripts to keep track of this all.
This isn't what's happening right now though as the httpd.conf is just ~9MB and MySQL is humming along happily. EDITED TO ADD THE FOLLOWING It should be added that 99% of the rehashes that lv-shared03 does in a day go through without issue and complete within a couple seconds, the problem is the random few that are taking > 60 seconds to run.
Francisco
Oh good to know got two in LV with you. I don't actively monitor them they're small and pointless.
Keep up the good work.