Page loading stalls

oscarBravo

I'm getting repeated "stalling" when loading pages the last day or two. I haven't seen anyone else mention it, but I'm not seeing it on any other websites, so it's hard to say whether it's a local problem. Maybe the fact that I'm one of the few accessing the site via IPv6?

I can attach a .har file that illustrates the problem, if it helps.

oscarBravo

Bump.

My suspicions about this being an IPv6 problem are increased by this post.

Boards.ie: Danny

Howdy,

Please attach a .har file so we can investigate. The office infrastructure is IPv4 only so if this is IPv6 related it might also explain why it's not being widely reported and make it that bit more difficult to debug.

Danny

oscarBravo

$ ping6 www.boards.ie
PING www.boards.ie(2a01:258:f000:3::108) 56 data bytes
64 bytes from 2a01:258:f000:3::108: icmp_seq=2 ttl=58 time=16.8 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=3 ttl=58 time=13.0 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=4 ttl=58 time=18.7 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=5 ttl=58 time=29.8 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=7 ttl=58 time=12.1 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=9 ttl=58 time=16.2 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=13 ttl=58 time=14.9 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=18 ttl=58 time=30.9 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=19 ttl=58 time=11.8 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=20 ttl=58 time=13.5 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=21 ttl=58 time=15.6 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=22 ttl=58 time=13.4 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=23 ttl=58 time=11.1 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=24 ttl=58 time=12.8 ms
64 bytes from 2a01:258:f000:3::108: icmp_seq=25 ttl=58 time=12.6 ms
^C
--- www.boards.ie ping statistics ---
25 packets transmitted, 15 received, 40% packet loss, time 24019ms
rtt min/avg/max/mdev = 11.177/16.272/30.947/5.886 ms

Tracing with mtr:

Host                                                    Loss%   Snt   Last   Avg  Best  Wrst StDev
 1. 2a02:3d8:5::1                                         0.0%    26    3.0   2.3   1.2   9.4   1.9
 2. 2a02:3d8:ffff:201::1                                  0.0%    26    4.1   2.0   1.4   4.7   1.0
 3. 2a02:3d8:ffff:200::1                                  0.0%    26    3.4  11.2   3.1  37.3  10.5
 4. 2a02:3d8::2                                           0.0%    26    6.2  13.0   5.6  41.4   9.3
 5. 2a02:3d8::1                                           0.0%    25    8.7  21.1   5.5 150.3  29.7
 6. 2001:7f8:18:12::22                                    0.0%    25   33.6  21.1  11.4  49.9  10.5
 7. 2a01:258:f000:3::107                                 41.7%    25   26.6  21.6  11.8  50.5  11.9

The attached (zipped) .har dates to the initial report, but I can probably capture a newer one if necessary.

Boards.ie: Danny

Cheers for that, given what's in the HAR file I've just removed it; I should have asked you to send it in over email

Danny

oscarBravo

Oh, cookies? Didn't think of that.

It's behaving itself now, just FYI.

Boards.ie: Danny

oscarBravo wrote: »

Oh, cookies? Didn't think of that.

Yep, me neither

oscarBravo wrote: »

It's behaving itself now, just FYI.

The packet loss is worrying, our sysadmins have restarted IPv6 on both load balancers. Please report back when/if it happens again and we'll try some other stuff

Danny

oscarBravo

I'm seeing intermittent low levels of packet loss on that last hop, but not enough to be a major disruption to my boards experience right now. I'm heading into a meeting for most of the afternoon, but I'll leave a trace running.

Cheers!

_Kaiser_

Performance here seems significantly improved in the last hour too (Win 8 Enterprise / Chrome)

oscarBravo

85,000-odd pings later, I'm seeing an average 3% packet loss between Digiweb's INEX node and boards.ie's IPv6 address. It's low enough that it doesn't impact noticeably on my browsing experience, but I thought you should be aware of it.