Morning all.
Bear with me, this is a very, very, very long post.
Never thought I would ever make a post that would make OVH look bad, but here we are.
I'll outline the problem a little bit just so that everyone gets the gist of things.
I am a patient person, so when my server was seemingly limited to 60Mbyte/s (~480mbit/s) back at the start of june (yes, twenty days ago, I said I was patient) I put in a support ticket after ensuring there was nothing at fault on my end.
Do note that if this was an intentional limitation by them, why wasn't it brought in line with the rest of the HG servers at 300mbit?
The server in question is an HG-XL (12x 3T disks) from august last year. This means that server is on a 40.0TB/month limit with unrestricted network/burst capability, and not 300mbit "unlimited" like the current line.
I chose to keep the 40T limit for favor of having the burst ability every now and again, and considering the box hardly did traffic aside from this I figured it would be left alone for a long time.
Anyway, onto the "support" I've received.
Note: I am a .ie customer, but that forum is as dead as OVH's customer support.
Initial ticket placed june 3rd.
Ticket outlined the issues, measures taken to ensure not a problem on my server (Able to run iperf tests at 6Gbit/s to iperf.ovh.net, and if I ran two tests concurrently to a friends 10Gbit server, it would max out my NIC.)
I have explicitly explained in this ticket that the problems are only OUTBOUND in nature (upload) and only on traffic LEAVING the OVH network.
Response was "please let us know if you're still having slow speeds" Answer was yes.
After that, I'm asked to run a wget test on the proof.ovh.net server which signals to me that support aren't reading customer tickets. I had previously explicitly explained that the issue is only OUTBOUND and only EXTERNAL.
None the less, I humor him and run the test.
Result is as expected:
Code:
100%[====================================================================>] 10,737,418,240 617M/s in 17s
2013-06-06 17:16:43 (620 MB/s) - `/dev/null' saved [10737418240/10737418240]
Again, internal was never an issue.
Responds saying to provide traceroutes, which I do to some single servers I had previously known to have good speeds (90-110MB/s FTP) and are now at 60.0.
Along with this, one of the other guys that uses a vnc gui (yes, I know, not exactly a great choice - but plenty of resources to go around) on the server likes to show off the speedtest.net results; but this afforded me with an opportunity to get these from the test history:
> Slow Uploads:
>
http://www.speedtest.net/result/2756418920.png
>
http://speedtest.net/result/2753709667.png
>
http://speedtest.net/result/2753708589.png
>
http://speedtest.net/result/2749553243.png
>
http://speedtest.net/result/2749462082.png
>
> This is the first speedtest with a slow outbound speed, Look at the date in the upper right corner:
>
http://speedtest.net/result/2749438739.png
>
>
> This is the LAST speedtest that had a high outbound speed, Notice the date in upper right corner, from two seperate servers within minutes of one another:
>
http://speedtest.net/result/2745251535.png
>
http://speedtest.net/result/2745249684.png
>
>
> These are a collection of fast outbound speeds:
>
http://speedtest.net/result/2721070933.png
>
http://speedtest.net/result/2705510277.png
>
http://speedtest.net/result/2699477369.png
>
http://speedtest.net/result/2400422784.png
>
http://speedtest.net/result/2369825066.png
>
http://speedtest.net/result/2369842910.png
3-4 days go by and I get a response saying that it appears to be resolved. It was not.
I had to say this: (copy/paste)
Hi.
The issue is ongoing and permanent.
These speeds are not intermittent, they are permanent, and have been exactly the same now for over one week.
Again, this issue is NOT resolved. External speeds are still only 60Mbyte/s.
The issue started when that ticket appeared. The issue has remained since that ticket appeared. That ticket changed something on the same network I am connected to (HG)
Issue not resolved. Issue remains the same for almost one week.
Ticket in question:
http://status.ovh.net/?do=details&id=4806
I should note that I only NOTICED the issue when that ticket appeared, It could have been going on for 4-5 days more (I remember seeing a burst speed on the 28th of 260MB/s)
He responded with:
It is possible that this is being caused by some maintenance we are carrying out on the network, but the issue shouldn't be lasting this long. If you would like to create an incident ticket regarding this issue, the technicians will have the ability to troubleshoot this issue further. You can create an incident ticket via your OVH Manager under Support > See My Incidents.
Now, I was apprehensive about putting in an Incident ticket considering how long it took me to get them to realize that the problem had nothing to do with inbound traffic, and figured they would just play coy and say not my problem not our fault!
Eventually I got told to bugger off (not their words), we can't help you any more.
This ends the saga of the support system.
Now begins the Incident support fiasco.
Ticket #1:
Explain the situation in the manner that their forms requested as per follows:
Section 1: Brief summary
I am unable to break 60.0Mbyte/s (~480mbits) when leaving the OVH network
Section 2: Slightly more info (I don't remember the exact words required as the ticket is ongoing)
I cannot break 60.0Mbyte/s if the destinations are outside
of the OVH network. Inbound speeds are fine.
Outbound has been at 60.0Mbyte/s for over two weeks now, I
was trying to resolve this with the other ticket system but
got nowhere.
I can run 10gbps tests internally (iperf.ovh.net, another
10gbps server at the same time = 1.1GByte/s)
This means it is not an issue with my server. My traffic
outside OVH is being shaped somewhere upstream.
This server is the old unlimited speed one, on a set data
usage per month. (40TB)
I have not requested to be changed over to the new
"unlimited" system with a speed limit.
This first happened between may 30th and june 3rd.
Section 3: "Extra" Info
I'm not exactly sure when the speed dropped as I never used
to monitor it this closely.
Please realize that I am talking about OUTBOUND (UPLOAD)
and EXTERNAL traffic only. It took several days for the
other support system to realize this.
External Inbound, Internal Inbound, Internal Upload are
FINE.
External UPLOAD is limited to 60Mbyte/s (480mbits)
I hope I have provided you with enough info to proceed.
At this point, I felt that maybe I would get this resolved.
This is the response I received a couple of hours later.
Dear customer,
I have just performed an iperf test between your server and
a test iperf server iperf.ovh.net :
-----------------------------------------------------------
-
Client connecting to iperf.ovh.net, TCP port 5001
TCP window size: 22.9 KByte (default)
-----------------------------------------------------------
-
[ 3] local 188.165.13.xxx port 47937 connected with
188.165.12.xxx port 5001
[ ID] Interval Transfer Bandwidth
[ 3] 0.0- 1.0 sec 695 MBytes 5.83 Gbits/sec
[ 3] 1.0- 2.0 sec 699 MBytes 5.86 Gbits/sec
[ 3] 2.0- 3.0 sec 700 MBytes 5.87 Gbits/sec
[ 3] 3.0- 4.0 sec 706 MBytes 5.92 Gbits/sec
[ 3] 4.0- 5.0 sec 207 MBytes 1.73 Gbits/sec
[ 3] 5.0- 6.0 sec 0.00 � ��s
147573952589675855872 Bytes/sec
[ 3] 6.0- 7.0 sec 0.00 � ��s
147573952589675855872 Bytes/sec
[ 3] 7.0- 8.0 sec 0.00 � ��s
147573952589675872256 Bytes/sec
[ 3] 8.0- 9.0 sec 0.00 � ��s
147573952589675855872 Bytes/sec
[ 3] 9.0-10.0 sec 0.00 � ��s
147573952589675855872 Bytes/sec
[ 3] 0.0-10.0 sec 2.94 GBytes 2.52 Gbits/sec
[ 3] MSS size 1448 bytes (MTU 1500 bytes, ethernet)
-----------------------------------------------------------
-
Client connecting to nsxxxxxxxxx.ovh.net, TCP port 5001
TCP window size: 21.9 KByte (default)
-----------------------------------------------------------
-
[ 3] local 188.165.12.xxx port 44437 connected with
188.165.13.xxx port 5001
[ 3] 0.0- 1.0 sec 610 MBytes 5.12 Gbits/sec
[ 3] 1.0- 2.0 sec 618 MBytes 5.18 Gbits/sec
[ 3] 2.0- 3.0 sec 620 MBytes 5.20 Gbits/sec
[ 3] 3.0- 4.0 sec 619 MBytes 5.19 Gbits/sec
[ 3] 4.0- 5.0 sec 618 MBytes 5.18 Gbits/sec
[ 3] 5.0- 6.0 sec 617 MBytes 5.18 Gbits/sec
[ 3] 6.0- 7.0 sec 613 MBytes 5.14 Gbits/sec
[ 3] 7.0- 8.0 sec 617 MBytes 5.18 Gbits/sec
[ 3] 8.0- 9.0 sec 618 MBytes 5.18 Gbits/sec
[ 3] 9.0-10.0 sec 619 MBytes 5.19 Gbits/sec
[ 3] 0.0-10.0 sec 2.02 GBytes 1.74 Gbits/sec
[ 3] MSS size 1448 bytes (MTU 1500 bytes, ethernet)
iperf:~#
I did not notice any issues.
Best regards,
employeename
I went a little irate in the next ticket as I was nearing the end of the rope over this, especially after that last response given the information I had provided.
Going to cut it up a little to make it neater, but every word of this (before "repeat") was on a seperate line.
I will spell this out one word at a time for you.
This Is Not An Internal Traffic Issue.
And the main issue:
This Is An Issue With Traffic Outside The OVH Network
repeat:
I CAN ACHIEVE CORRECT SPEEDS INSIDE OVH.
I CANNOT BREAK 60.0Mbyte/s OUTSIDE OVH NETWORK.
(~480Mbit/s)
This is NOT a problem internally. iperf.ovh.net is
internal. I provided this information in the first ticket.
Please read the first ticket in its entirety.
After that, I had some friends with servers in various places around europe (including OVH) do speed tests using a test file sitting on a ramdisk on my server
.uk: 100%[================================================== =============================>] 1,073,741,824 55.9M/s in 18s
.de: 100%[================================================== =============================>] 1,073,741,824 57.0M/s in 20s
.ch: 100%[================================================== =============================>] 1,073,741,824 51.4M/s in 23s
.uk: 100%[================================================== =============================>] 1,073,741,824 45.9M/s in 22s
.ovh: 100%[================================================== ============================>] 10,737,418,240 654M/s in 16s
UK1 = vpsnet
UK2 = Linode
Not sure where the others are hosted, but I will link him this.
Yes, that's a 10GB file, on a 10Gbit server in OVH's datacentre.
Anyway, after all of this was provided to the clown support team, I get this in a ticket:
Now, I would understand asking me to change kernels (I did it anyway, result is the same) if my OVH -> OVH speeds were slow, but they're not.
And here we stand. I've requested a new tech and we'll see if they are any good.