Too many troubles and weird issues, probably leaving jaguar.

[Himeko]

Hi, been a customer of you guys for a few months now, even brought a friend in for another VPS, but lately I'm seeing things I dont like one bit.

jaguar staff blocked 2 sites of mine and 1 of my friend, 2 of them saying they had high load and were causing issues and trouble for other nodes.

Well, to my understanding for years, load issues on a VPS are isolated to that node and CPU share etc is limited, so why would it cause any trouble to other customers?

The staff modified the .htaccess file with a "deny all" to block access, and also modified the file permissions so it cant be modified or renamed, besides setting the user etc to root. We as vps admins can go through SSH and remove the attribute so that we can edit and rename them, so what's the point in doing that? just giving us extra trouble.

But despide editing the file and removing the "deny all", the file is sorta cached or something, cause access remains blocked unless we rename the file to something else.

This has become a very annoying issue for the last several weeks and something we think completely overboard for your staff to do on a vps/dedicated machine on their own.

[Gwaihir]

VPS offers a lot more isolation than standard shared, but nevertheless a highly stressed VPS can still take so many limited resources that it impacts other VPSses on the same machine.

I don't know why it's handled like that. Perhaps to give a clear signal that you need to deal with the underlying issue first, rather than just reenable it?

[JPC-Bilal]

Hi Himeko,

We are sorry for the inconvenience. We usually warn our clients first if we find their sites are causing load but sometimes we have to do it if a VPS starts causing load on the node and if it starts effecting other clients. We try to provide as much information as possible so that clients can fix the issues with the sites. Once they do we enable them again.

Can you please provide us the Ticket# in which you were contacted regarding this so I can check and update you. Thank you.

[Dayron]

I'm on the other end of Jaguar's resource mismanagement. We seem to have a rogue VPS on our physical node that brings my VPS to a stop. I have noticed before and reported it and Jaguar was dismissive saying they couldn't see anything wrong. However, recently a client of mine began HEAVILY relying on their web service. Need less to say I have gotten multiple calls from them from 9/9/2011 to today (9/26). Its always the same story. All of a sudden their webapp (a small instance of Zen Cart) becomes totally unresponsive. Then their website stop reponding. I try other client sites...no dice. I contact Jaguar. Response has been the same since 9/9... a VPS node was disrupting all the others and they have corrected it. I guess it's been 3 or 4 times now this has happened. So I installed Munin so I could see if those outages show up at times i'm not aware of it. Lo and behold yes it happened sometime before my client's office hours as well. So Jaguar needs to figure out how to properly manage their allotment of VPS resources or be forthcoming to those who buy into their service that "other nodes can cause your node to become unavaialable at any time".

Here is a link to an image from Munin showing how my Monday uptimes did today.

I'm on the other end of Jaguar's resource mismanagement. We seem to have a rogue VPS on our physical node that brings my VPS to a stop. I have noticed before and reported it and Jaguar was dismissive saying they couldn't see anything wrong. However, recently a client of mine began HEAVILY relying on their web service. Need less to say I have gotten multiple calls from them from 9/9/2011 to today (9/26). Its always the same story. All of a sudden their webapp (a small instance of Zen Cart) becomes totally unresponsive. Then their website stop reponding. I try other client sites...no dice. I contact Jaguar. Response has been the same since 9/9... a VPS node was disrupting all the others and they have corrected it. I guess it's been 3 or 4 times now this has happened. So I installed Munin so I could see if those outages show up at times i'm not aware of it. Lo and behold yes it happened sometime before my client's office hours as well. So Jaguar needs to figure out how to properly manage their allotment of VPS resources or be forthcoming to those who buy into their service that "other nodes can cause your node to become unavaialable at any time".

Attached is an image I posted to google docs showing how my Monday uptimes did today. Also the link to the image is below.

https://docs.google.com/a/redhourtec...xYzkz&hl=en_US

[JPC-Bilal]

Hi Dayron,

Do you have a ticket opened? If yes please provide us its number and if no, then please open a ticket so that we can look into it. Thank you.

[Dayron]

ticket #13435309. The ticket is open and has been since 9/9 and drops continue and as typical the most recent reply this eve summerized was the typical response "we can't see anything wrong right now" and got some output regarding recent server load averages etc. I'm not asking the tech to defend the server. I'm just asking the tech to find the issue and stop it or tell me what i can do if it's on my end. But thus far these kinds of outages all seem to be related to some other VPS hogging resources. If migrating to another service is the best option to prevent this let me know.

Hi Dayron,

Do you have a ticket opened? If yes please provide us its number and if no, then please open a ticket so that we can look into it. Thank you.

[JPC-Bilal]

I am checking that ticket and will update you shortly in the ticket. Please follow up there. Thank you.

[Dayron]

I am checking that ticket and will update you shortly in the ticket. Please follow up there. Thank you.

I've reviewed this ticket since I opened it and it is becoming clearer to me that Jaguar has no ability to prevent a VPS from killing other VPSs. They can react only and hope and pray it doesn’t happen again. So here is the full-meal-deal on this subject. It's long, but if you want to know exactly what to expect as someone reviewing Jaguar's VPS as an option consider the info below. Also, hopefully Jaguar will take this seriously enough to make the changes needed to prevent this nonsense of one customer able to hinder another customer's operations.

9/9/2011 11:27
I report to Jaguar my server is unavailable. Only 25 minutes later a tech replies (to Jaguar's credit their techs tend to reply quickly...grade A on that). The tech says there were some load spikes causing the problem but looked like things were normal again could I confirm. One thing that bugged me was they didn't seem interested at first to tell me what the issue was or totally ID the problem. I expressed concern and asked what I was running that caused the issue or if I could block an IP address to prevent the issue. I was under the impression the issue was on my VPS and I wanted to do what it took to resolve it.

12:52 the ticket was raised to Level 2

1:12PM same day - Jordan responded giving me the details I requested. I was glad it wasn’t me, but a little surprised at the info provided in the reply:
"The load was on the actual hardware node and not on your specific container. Our administration has been monitoring this issue and will continue for some time. The offending container appears to have been taken care of and the load is stable on the server now. If you do happen to notice any other problems, please don't hesitate to let us know".
I figured this was the end and went on my way.

09/12/2011 10:55AM
I reported HTML seemed to be responding ok, but MySQL response was horribly slow and ran some code to display MySQL query times... 66 seconds was its response at that moment...VERY slow.

12:05 - Jeff responds that he is passing on to the admin for further investigation similar to the reply I got previously where the administrator was monitoring the situation. I feel like I am growing ever closer to the Death Star at this point... "I've got a bad feeling about this".

2:20PM same day I get a response very similar to the response I got when I reported the problem on 9/9. It was somewhat dismissive. The reply was from Sajjad and is as follows:

"We are sorry to hear that you are facing this problem. I have checked the server uptime and Apache service uptime show that there was no recent issue on the server. Things are working normally and we are not having any network issue too. The load and traffic is normal on the server and site is loading fine without any issues. I have verified it from my side. I am sorry but I was not able to reproduce the problem from my side. Can you please check again and if problem persists perform a traceroute to this site from your side and provide us with the results to see where the problem lies. However we noticed some load spikes on your VPS earlier today due to which you might have experienced slow response from the server. Thanks for the cooperation. Feel free to contact us if you need further assistance."

So it seems in these cases the Jaguar standard level 1 response is "Its working now...bye". It seems you have to beg for them to take it up a level. So needless to say I replied at 2:44PM:

"During the slow periods I logged into WHM (though very slow). I reviewed running processes and there was no indication of heavy load. Were the load spokes on my VPS occurring during the times of reported slowness? Also could you tell me the exact times of the spikes and what service was causing the problem?"

Again I wanted to be sure that this time it wasn't my VPS causing the hangups and wanted to do whatever it took. However, given the next reply I realized this is a Jaguar service problem:

3:18PM same day Chris Sweeney replied:
"It appears that the problem was on another VPS on the node you are on. With the way that VPS's are setup if a problem on the node exists, it will spread to the other VPS's on the node. If that VPS crashes, it won't affect you as they are in their own operating environment, but if they are using a lot of processor and memory, it will be felt across the entire node. We do monitor for these issues and respond to them. We disable the content/user that is utilizing the resources and we then notify them so they can fix this. We apologize for the inconvenience."

Great. Its the same issue. The tech states they monitor for these issues, but other techs dismiss it with "its working now" when it happens again and I have to ask for more information each time and it seems the issue has been the same in just a few days. My hope was that they finally have corrected the issue and that was that. But no.

9/26/2011
A client called me at 4PM saying they were unable to access their website and their e-store. The symptoms were exactly the same. I was tied up and unable to report the ticket right away, but I knew what was coming when I did. I was going to be dismissed at first. Then I was going to have to ask it go further. Then it was going to be that another VPS was causing all the trouble and they took care of it. But, I updated this ongoing ticket anyway and reported the trouble.
At 9:40PM i was finally able to update the ticket I explained the issue. Also requested the look at the Munin graph (Installed Munin after this happened the second time) that showed the times the server had become unresponsive that day (to my surprise it was 3 times though only reported once). Guess what I got in response?

9:46PM same day - Anoop Mohan replied right away. Again I'm usually very happy with Jaguar repose time. But it seems these days a quick response means I get dismissed quickly. He gave the same line in summary of "everything is fine now". He did go as far to show resource usage of a period of time etc. But in the end he said this:
"At the moment I couldn't find any issues with your vps/node. Kindly let us know if there is anything else we can do for you." He also asked for a link to the Munin graph I referred to. I suppose he was too busy to login to the server and check like I had requested. So I sent him the screenshot which is attached in an earlier post in this thread and I have attached again. All other munin graphs look the same. You see service...then drop outs all at the same time.

Keep in mind from 9/9 to now this is all one ticket and every tech that is the first responder seems to act like they can't read the posts previously. It gets treated like a new issue every time. Frustrating to say the least.

09/27/2011 12:53AM
Bilal B responds with a similar song sung to the same tune:
"Thank you for your patience with this issue. I have looked into it and investigated it on my side. I have checked the load history of node and found that there were couple of high load spikes on the node around the time you mentioned above. Probably these load spikes cause your VPS and sites to become slow and inaccessible though these spikes remained for short period of time". He also assures me they will monitor the server (as I had been assured before) and will take care of the offending VPS. That was to seen.

05:54 AM
Something different at last! Bilal B sends a response saying they are still monitoring though they haven't seen the spikes. While the issue wasn't ID'd there the fact I got a proactive response that seemed to indicate they really were monitoring the server this time was a shift in the right direction I thought. The outcome of that shift was still to be determined.

09:17AM
Catalin S responds with this:
"We have identified a vps that was causing issues on the hardware node and disable it. Please let us know if you still notice slow loads".
I'm glad they ID'd the problem-container, but...didn't they so that on 9/9? I guess it wasn’t clear if they actually found the issue on 9/12. Then as said above they ID'd and disabled the offending VPS...again.

This approach is reactionary only. It seems they closely monitor the situation only after someone reports an issue. At that time the person (me in this case) has to continue to hammer on it and prove to tech support there is actual trouble. It has taken 17 days (up to now) to hopefully resolve the issue at hand. But this does not address the ongoing issue of one VPS killing others. I suspect their Hybrid plan may have been in response to this fiasco. They seem to admit that "at times" there maybe 40+ VPSs on a server (a passive admission perhaps that Jag WAY over-sells their servers) and for quite a bit more a month you can ensure you aren't on a server that’s oversold. But even then on the Hybrid plan can Jaguar ensure that one of the few VPSs on that server won't bring the others to a standstill?

All this to say... Jaguar, get proactive, ID the issues that do occur and take measures to prevent them before they occur. If you know one VPS can wreck the others then configure it so it can’t do that.
I've been a loyal customer for years even through some crappy times at Jaguar. There has been a nice stretch of stability that now seems to be disrupted by other customers?? I can only suspect since this just started happening recently that Jaguar has oversold the physical server my VPS is on. I hope that’s not the case either. Either way it needs to get fixed.

Graph from 9/26/2011
jag-off.png

[JPC-Nazmy]

I have investigated your vps settings and I think there was a misconfig on some parameters regarding memory and TCP traffic , I have now changed this settings and urge you to test performance for 48 hours and update your ticket accordingly .

[Dayron]

I have investigated your vps settings and I think there was a misconfig on some parameters regarding memory and TCP traffic , I have now changed this settings and urge you to test performance for 48 hours and update your ticket accordingly .

Ok so it WAS another VPS server causing all problems in my 3 reports and that was stated by 3 techs (including yourself). But now its a memory and TCP configuration issue... on my VPS specifically, the Virtuozzo platform, or the physical server? My issues don't occur at 48 hour increments most of the time. it may be a week between issues. So the next 48 may be wonderul and in 72 hours it all hangs up again. This goes back to Jaguar taking a reactionary approach. Monitr it for the next 2 weeks maybe... or monitor the services Jaguar offers to ensure they are behaving has advertised.
On 9/12 at 3:18 PM Chris Sweeney told me this:
"We do monitor for these issues and respond to them".
So if Chris is correct you should be monitoring my server for 48 hours, 48 days, 480 days, etc, for the life of my service. My requestto you is monitor my (and everyone else's) servers the way we are told they are monitored so Jaguar can respond to an issue perhaps before the reseller or the their client's even know there is an issue then tell us theissue was resolved before we knew it. That would be wonderful! I've had to post and update tickets and now take it to the forums where maybe this gets resolved for good for everyone.

[Dayron]

I have investigated your vps settings and I think there was a misconfig on some parameters regarding memory and TCP traffic , I have now changed this settings and urge you to test performance for 48 hours and update your ticket accordingly .

Looks like the memory and TCP traffic misconfiguration still exisits or that wasnt the issue at all. Is the physical node simply oversold? I'm having the mysterious slowdowns again. So far Jaguar has not responded to me informing me of them. This is further evidence they DO NOT monitor the servers. Any chance this can get resolved before we hit a full month of this?

I guess the question now is how can this get taken above the heads of the techs that cant fix the problem? What is the next step?

uptime-day.png

[Ron]

I'd give up and ask to be moved to a different server. See if that fixes it for you.

[Dayron]

I'd give up and ask to be moved to a different server. See if that fixes it for you.

I've ocnsidered making that request. But all the answers they give me (when I demand an answer) always seem so "Ok we fixed it". However, I'm starting to feel like a victim of the Jedi Mind trick. If their approach to server migration is anything like their approach to server monitoring and this issue resolution I'm scared of the prospect. I may be getting used to limping on a broken foot. I don't know if I want to do anything to risk breaking the other one.

[JPC-Zishan]

I am sorry for your issues. I have your ticket claimed and I am working on it. Will update you shortly with my findings.

[gohighvoltage]

Hey All, sometimes technical problems can be a bear to pinpoint and repair. Give Jag some time, they will get it corrected and you will he happy. They have never let me down, and there service has been way better then other companies I have dealt with.

As far as Himeko, before you leave Jag, look around the internet. There is hundreds of stories, from tons of other hosting companies, of people getting shared and VPS account shut down due to extreme load that disrupts other clients.

If your load is that high, it is time to move on to a dedicated server. Nobody can expect a hosting company to allow one client to disrupt service to multiple other clients.

I read so many rants about other hosting companies shutting peoples accounts down, but all of the hosting companies pointed out that they need a dedicated server.

If your getting that much traffic on your sites, then it only makes sense that you would want a dedicated server.

Good Luck my friends. Take Care.