Some progress on the forum issues

Hi, all -

Just a quick note on where things are at with the forums and the odd/
frustrating issues we’ve been seeing over the past several months. I
spent the afternoon with the folks in Provo going over the issues and
finding out where things are at and what the plan is moving forward so we
could have a clearer picture of what’s happening.

  1. The logout issue should be fixed now. The page does redirect to a
    Novell-branded page, but for the time being, I’m considering this a win -
    we’ll get it corrected properly in the near future, but there should be
    no 404 errors on logout.

  2. There have been some patches applied to the authentication servers
    that seem to have stabilized the majority of the logout issues. There’s
    still one issue that we’re trying to nail down where a logout happens
    after an unexpectedly short time (say 10 minutes from login). That
    appears to be unrelated to the actual authentication server issues, as
    it’s related to the cookie set by vBulletin. It appears (we’re not 100%
    sure yet) that it might be tied to a Javascript issue in vBulletin; this
    ended up being reproduced by chance while kgroneman and I were meeting
    with the guy who’s been working on the issue, and he noticed immediately
    before that a Javascript triggered by clicking on “reply to thread”
    seemed to hang up.

When he refreshed the page, his cookie had been expired and he was logged
out. Clicking on the “Login” link, he was immediately logged back in
(without a password request) because the authentication server session
was still active (so there’s no security issue there - you had to
previously have logged in with your ID in order for the login to pass-
through like that - and that’s by design).

Currently we’re looking to see about increasing the vBulletin session
timeout. We /may/ also be able to increase the overall forum session
timeout, but first we want to get things stabilized without tweaking too
many variables.

There also was some discussion about some additional configuration
changes on the authentication boxes - memory in particular was
discussed. The new auth servers are 64-bit - whereas the old iChain
boxes were 32-bit (as iChain was only available as a 32-bit product), so
the new setup can handle a lot more memory, which should help. There
were other optimizations that they talked about and are going to look
into as well.

As for specific things that you can do to help us:

  1. If you run into a problem, we need to know. If you can outline steps
    to reproduce it as well as the time and date you had the problem, that
    will help us determine if it’s before or after a particular change was
    made.

  2. Headers/traces can help. HTTPFox (for those using Firefox) can
    provide some information here. If someone knows of a Chrome plugin that
    has similar functionality, let us know, because we want to also test with
    Chrome as well as Firefox and IE (for those few Windows users we get
    around here - which aren’t many). tcpdump or wireshark info can also be
    useful, but often provides a lot more info than is needed - so filtering
    to HTTP/HTTPS requests would be ideal (if you grab HTTPS, make sure you
    are using some sort of proxy with its own private key to decode the trace
    with - and you’ll want to obscure any auth info in the trace as well - if
    you don’t know how to do this, don’t worry about getting HTTPS traces -
    HTTP headers are the most useful for us).

  3. The more detail you can provide on what page you started from, what
    you clicked, and so on - the more useful your report is to us.
    Especially if you can reproduce the issue entirely at will.

I know I’ve said it before, but I have to say it again: Thanks again to
everyone for your patience. We know these issues have been very
frustrating to deal with, and it has taken far longer than we anticipated
to get to this point.

I am feeling optimistic that we’re nearing the end of these issues - we
certainly are closer today to the end than to the start.

Jim

Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

> we certainly are closer today to the end than to the start.

thank you for your caring, hands on involvement…


dd

On Wed, 23 May 2012 00:10:47 +0000, Jim Henderson wrote:

> 2. Headers/traces can help. HTTPFox (for those using Firefox) can
> provide some information here. If someone knows of a Chrome plugin that
> has similar functionality, let us know, because we want to also test
> with Chrome as well as Firefox and IE (for those few Windows users we
> get around here - which aren’t many). tcpdump or wireshark info can
> also be useful, but often provides a lot more info than is needed - so
> filtering to HTTP/HTTPS requests would be ideal (if you grab HTTPS, make
> sure you are using some sort of proxy with its own private key to decode
> the trace with - and you’ll want to obscure any auth info in the trace
> as well - if you don’t know how to do this, don’t worry about getting
> HTTPS traces - HTTP headers are the most useful for us).

For chrome:

ctrl+shift+I -> network

That will give us similar information to what httpfox provides. :slight_smile:

Jim


Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

Another thing that I should mention as well - over the past several
months/years(?) since the new editor came in on the web interface, we’ve
had many people ask if the advanced editor could be enabled by default so
the code tag button is there.

A recent update introduced a configuration setting to enable this - so we
have enabled it now. There should not be a need to go select the
advanced editor any more - it’s been set to be the default. :slight_smile:

Jim

Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

On Wed, 23 May 2012 16:08:04 +0000, Jim Henderson wrote:

> A recent update introduced a configuration setting to enable this - so
> we have enabled it now. There should not be a need to go select the
> advanced editor any more - it’s been set to be the default. :slight_smile:

Note that this change is not in the quick editor, but when you reply and
used to get a “basic editor”, now you get the “advanced editor”. The
quick reply box (by its nature) doesn’t have that kind of functionality.

Jim


Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

On Wed, 23 May 2012 00:10:47 +0000, Jim Henderson wrote:

> 1. The logout issue should be fixed now. The page does redirect to a
> Novell-branded page, but for the time being, I’m considering this a win
> - we’ll get it corrected properly in the near future, but there should
> be no 404 errors on logout.

Slight change on this one, it seems that while the logout is being
processed, the cookie from vBulletin isn’t yet being cleaned up - the
tech team is aware and looking into that. So progress, but not 100% done
yet.

Jim


Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

Has this really been enabled ? It doesn’t work for me either clicking “reply” or “reply with quote buttons” under the post. It doesn’t redirect me to the advanced editor if I push the “reply to thread” button either.

Best regards,
Greg

I get it with “reply to thread”, but in my view (I use the hybrid view), I see what you see. Will have to see if that’s expected - it was my understanding that the config change did enable this, but I don’t use the web interface often.

Which display view are you using?

Jim

Did a little more digging, and it seems that Kim and I had a failure to communicate. He set it so the WYSIWYG editor is the default, but that editor doesn’t have the same buttons as the advanced toolbar.

Back to the drawing board - shoot, I thought we had that one sorted. Sorry for the false alarm.

Jim

No problem :slight_smile: using linear mode btw. Here’s a screenshot of what I see when pushing “reply with quote” :
SUSE Paste

hendersj wrote:

>
> hendersj;2464983 Wrote:
>> I get it with “reply to thread”, but in my view (I use the hybrid view),
>> I see what you see. Will have to see if that’s expected - it was my
>> understanding that the config change did enable this, but I don’t use
>> the web interface often.
>>
>
> Did a little more digging, and it seems that Kim and I had a failure to
> communicate. He set it so the WYSIWYG editor is the default, but that
> editor doesn’t have the same buttons as the advanced toolbar.
>
> Back to the drawing board - shoot, I thought we had that one sorted.
> Sorry for the false alarm.
>
> Jim
>
>
> –
> Jim Henderson
> openSUSE Forums Administrator
> Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C
> ------------------------------------------------------------------------
> hendersj’s Profile: http://forums.opensuse.org/member.php?userid=5442
> View this thread: http://forums.opensuse.org/showthread.php?t=475513

As far as I understand Jim. This required some tinkering that Kim wasn’t
prepared to do.
But I see other forums using vB like us and the feature is in use there.

On Fri, 25 May 2012 03:41:26 +0000, caf4926 wrote:

> As far as I understand Jim. This required some tinkering that Kim wasn’t
> prepared to do.
> But I see other forums using vB like us and the feature is in use there.

Yeah, I did a little digging as well and it seems that there’s a style/
template tweak that could do this, but most of what I found was for
versions older than we’re running, so the instructions don’t translate.

Jim


Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

On Fri, 25 May 2012 03:36:02 +0000, glistwan wrote:

> No problem :slight_smile: using linear mode btw. Here’s a screenshot of what I see
> when pushing “reply with quote”

Yep, that’s consistent with what I saw. Not sure why I got the advanced
editor now.

Jim


Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

On 05/23/2012 02:10 AM, Jim Henderson wrote:
>
> I am feeling optimistic that we’re nearing the end of these issues - we
> certainly are closer today to the end than to the start.

and today i join you in optimism!

for today i discover i can connect to both (for example):


[someone's] Profile: http://forums.opensuse.org/member.php?userid=nnnnn
View this thread: http://forums.opensuse.org/showthread.php?t=nnnnnn

AND once on someone’s profile page i can click on “Find latest posts”
and “Find latest started threads” ALL with just a very brief (and
very acceptable) moments (one or three second, or so) delay!

THIS is progress: please ask the powers that be to snapshot and backup
the system and never “upgrade” again… :slight_smile:

(and, security patch only after plenty of testing)

ymmv


dd

On 07/04/2012 11:35 AM, dd@home.dk wrote:
> THIS is progress

well, it wasn’t progress at all, instead just a temporary change in
condition…

because, today we (maybe just me?) are right back to three minutes (or
so) getting to (for example)
http://forums.opensuse.org/member.php?userid=5442 or
http://forums.opensuse.org/showthread.php?t=475513

causes me to wonder if the recovery from
suse.com/suse.de/novell.com/opensuse.org all being out (yesterday)
included a restore from backup which included vBulletin configuration
files containing the molasses.

oh wow, and now that the page has come up, i see that the html/css is
wrong in that the page layout is nothing like it is usually:

1 http://paste.opensuse.org/10746195
2 http://paste.opensuse.org/4059382
3 http://paste.opensuse.org/43167217
4 http://paste.opensuse.org/19076385

WOW, it is even giving the WRONG user! (was supposed to be “hendersj”
and part of the page shows that, but it then lists the posts of
“stamostolias”)

i wonder if “vBSEO 3.5.2” is correct…


dd

On Mon, 09 Jul 2012 10:01:52 +0000, dd wrote:

> causes me to wonder if the recovery from
> suse.com/suse.de/novell.com/opensuse.org all being out (yesterday)
> included a restore from backup which included vBulletin configuration
> files containing the molasses.

Unlikely, as the outage affected more than the forums, it was probably an
infrastructure issue rather than a server issue.

I’m not sure that speculation about why things are now the way they are
is helpful. Facts and data are helpful. Speculation just spreads/drives
FUD.

And I know how much you hate FUD. :wink:

Jim

Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

On Mon, 09 Jul 2012 18:55:22 +0000, Jim Henderson wrote:

> Unlikely, as the outage affected more than the forums, it was probably
> an infrastructure issue rather than a server issue.

Just confirmed, it was - an odd and highly improbable cluster of proxies
got knocked offline. Nothing to do with any of the actual servers.

I have passed along your report of slow performance.

Jim


Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

On 07/09/2012 08:59 PM, Jim Henderson wrote:
> I have passed along your report of slow performance.

is no one else seeing the apparent html or css or db problems also reported?


dd

On Mon, 09 Jul 2012 10:01:52 +0000, dd wrote:

> WOW, it is even giving the WRONG user! (was supposed to be “hendersj”
> and part of the page shows that, but it then lists the posts of
> “stamostolias”)

Stamos is a friend according to the database, and the new page shows not
just the selected user’s activity but their friends’ activity as well
when you select “all”.

So that is actually behaving as expected.

Jim

Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C

On Mon, 09 Jul 2012 19:41:46 +0000, dd wrote:

> is no one else seeing the apparent html or css or db problems also
> reported?

I can’t say nobody’s seeing it, but it looks OK to me.

Jim


Jim Henderson
openSUSE Forums Administrator
Forum Use Terms & Conditions at http://tinyurl.com/openSUSE-T-C