i notice *lots* of posts regarding failed pings, but i don't see any actual help.

there must be experienced users who have seen this problem.

i've been using MT for a few years. somewhere around version 2.4-2.5, pings to websites just stopped working. creating new entries no longer causes the blog to appear on weblogs.com or blo.gs.

no biggie, but i just hate it when things don't work. and i see in forums that many other people are having this same problem.

are the forums just questions with no answers? i always post answers when i have them (to forums in general), but i don't see that happening a lot here.

I know you're frustrated, but these forums are filled with volunteers - if you want an answer, you need to be both specific about the problem that you're having, and patient. A search of the forums might reveal that I've answered several similar questions in the last week

Now, when you say the pings are failing, are you getting a "ping timeout error"? Have you tried going to weblogs and checking to see if your site is listed? You might want to go into your mt.cfg file and increase the "ping timeout" to 60 - this seems to take care of the error messages for all pinged sites except Wanderlust, but even then, I get the error, and my site is still showing up on the WL list of recently updated sites. If it's showing up, I'm okay with the error message, and it's my understanding that it's not so easily "fixed" because it's an issue with how the receiving sites respond to being pinged.

If it's not a ping timeout error, what error message are you getting?

Donna

yes i am checking if the ping is actually working on those sites (weblogs.com, blo.gs). no, there is no timeout error when i submit the entry. the ping screen appears briefly indicating it is pinging the various sites, and then brings me to the "rebuild" popup.

everything behaves as it did before, the only difference being that my site wasn't actually updated on the weblog sites.

i don't see any "ping timeout" var in the mt.cfg file.

and that's about it.

thanks for responding. i appreciate all help.

If you've been a longtime user of MT, you'll want to download a new installation and check out the mt.cfg file - changes are made with new releases that are not captured in the upgrade versions (because a new .cfg file would overwrite user paths and customizations). That's how I found "ping timeout":

I noticed this as well:


You might be a rare case

Donna

You can also just read the manual on configuration settings to see the things you can use in mt.cfg.

Or read about them on the MT Wiki. (I tried to organize them into more coherent groups there.)

well i tried
PingTimeout 60

and that didn't fix it. in fact before and after setting that var, the "ping screen" only displays for about a second, so i don't think timeout is really an issue.

my isp has my host configured with 27 different ethernet interfaces. yeah, i know, i couldn't believe it either. who could ever need more than one! (jk)

so i tried out "PingInterface" var with several different of these interfaces.

i noticed that these interfaces are configured with 2 different broadcast addresses, so i made sure to test with ips from each group.

still nothing.

but the more i think about it, the more i think it must have to do with the networking on my ISP rather than changes in MT. if it were some MT issue like bad RSS-Ping urls, more people would be complaining. it's likely a change that my ISP made that MT doesn't take into consideration.

in any case, it most definitely does not work.

anyone know what type of request this is? they call it a ping, but a udp ping can't carry along info. i'm thinking it's more of an HTTP GET/POST. if i had root access on my isp, i'd run a sniffer and debug it

i guess i could grep the code for the section and debug it myself. but i'm a bit afraid to look at the code for fear that i'll have the same reaction i did when i looked at bugzilla code, "OMFG, i can't believe this even works", after which i promptly discontinued use of bugzilla (which i've heard is being rewritten). also, i'm a bit lazy for that. was hoping someone who's seen the problem has figured it out...

ok, i can't help myself. i'm going to try to debug it myself.

just wanted to point out that after looking at the code, i have to say it's quite nice. methods are all separated in an orderly fashion into separate module files. code is actually readable (a miracle considering it's perl). logic looks logical he's even got embedded documentation for stuff that only he's likely to look at. indications that ben (i think that's his name) knows what he's doing. and, my personal obsession, all code is properly nested

that's re-assuring. and i can now also assume that ben wasn't involved in development of bugzilla.

nice work ben.

OK, I have enough info to strongly believe I've found a real bug.

The problem is that a new log entry fails to properly update weblogs.com, blo.gs, and moveabletype.org.

I added a print line to lib/MT/XMLRPC.pm and found what MT is sending to weblogs.com.

I then manualy did a GET (using the perl GET script available with most perl installs) and this is what I saw:




##################

everything looks good. I checked the weblogs.com site, and indeed, my site *was* updated.

but when i add an entry via MT, none of the sites are updated.

i am running 2.64. i got paranoid that maybe the update didn't install the latest MT lib files, so i downloaded the full version and diff'd the dirs. no diffs (except for my printf lines).

i wondered whether it was mis-scoped vars that caused problems when more than one site was pinged. so i turned off blo.gs and weblogs.com and just pinged moveabletype.org. but no luck, it still wasn't updated with my blog.

my current suspect is LWP.pm.

the version my ISP is running is:

# $Id: LWP.pm,v 1.84 1998/11/19 22:04:48 aas Exp $

don't know the latest version, but i'm sure this is not it. however, MT did work properly a while back (first noticed it didn't work about a year ago or so).

so my level of debugging has exceeded my interest in seeing this fixed. i will however submit any info you'd like about versions of stuff. and i will help debug this issue if you want to provide test libs, etc...

harry

OK, collected a bit more info on this and submitted a bug report. we'll see if they're interested in fixing this.

I've merged the original topic with the bug report topic, to keep everything together.

And yes, Ben is good. Very good.

well, judging from the threads here in the bugs section, seems like i might have more success with a thread in the general section where another user who has seen the same problem may have a workaround.

it just really bothers me when something, anything, just suddenly stops working for no obvious reason. man that's an annoying peeve to posess in such a modern (ie dysfunctional) world as this.

well, seems this isn't so much a bug reporting forum as a black hole for complaints

but i do have a follow up on my "bug"

i never realised that certain failures were logged. i always assumed that post, ping, other failures would be indicated in the windows that follows an action.

anyway, my 'activity log' clearly indicates the problems with the pings. i see two types of log entries:

1)

Ping 'http://www.movabletype.org/update/' failed: HTTP error: 501 Too many arguments for substr

2)

Ping 'http://www.movabletype.org/update/' failed: HTTP error: 501 Can't locate object method "new" via package "LWP::Protocol::http"

the second one is obviously a perl error. i guess one of the previous upgrades contained a new perl dependency.

the first error, i'm not clear what that would be, but i'm thinking it might have something to do with the missing perl package. so i'm going to get the proper package and see if all problems go away.

are there any release notes with each release? i looked and didn't see any. there should be such notes, and they should clearly indicate any new dependencies and such.

anyway. going to fix this now

I think I'm having the same problems as you... I cannot ping weblogs.com I get this error...



I've tried everythingh to get it working but nothing.

Although I can now ping bl.gs, I had to physically add it to their list thru their website. And now I appear on there.

So either weblogs.com has a fault, which I doubt, or MT has a bug when submitting to weblogs.... I think the latter.

I have a feeling this is the same problem a-drive posted about.

Ping 'http://rpc.weblogs.com/RPC2' failed: HTTP error: 500 read timeout

That's the error that shows up in my activity log after I published a new post. This is my first day on MT so any ideas on how I could keep that from happening would be great.

That's three very different problems. hlslaughter's got an old Perl module, a-drive's getting a Frontier error from weblogs.com (Frontier being the UserLand operating system that runs it), and David's getting a read timeout because weblogs.com isn't replying before his copy of MT gets tired of waiting. I see where Dave Winer said earlier today that weblogs.com is only going to re-render the HTML for the page once every five minutes, so it's a safe guess that the server's getting overloaded. Again. You can cut down the number of timeouts by adding

PingTimeout 60

to your mt.cfg file (going from 30 to 60 took me from 2/3 timeouts to more like 1/10 a couple of weeks ago), as mentioned at the top of this thread.

I am getting these exact errors when trying to ping anything in MT. This is with a clean install of the MT 2.64 with libraries distribution. Could there be an out of date file in that distribution? I did notice that the SOAP directory under extlib in that install was a little munged up (mt-check didn't recognize SOAP::Lite as being installed).

Anyone got any help on my error?

on my isp, /usr/bin/perl points to perl 5.004

/usr/bin/perl5 points to 5.6.1

i changed the .cgi files to point to the latter version of perl, and everything works again.

don't know why it worked before and then stopped.

but i can now spam weblogs.com again (JK).

thanks for your replies

try the same thing i did, make sure the version of perl your cgi scripts point to is a relatively new version.

Thanks, hlslaughter. Pointing to perl5.6.1 did the trick for me. From some reason 'whereis perl' wasn't turning it up but it was in the /usr/bin dir when I listed that directory.

How would I go about doing this hlslaughter? Kinda new to all this CGI biz...

Any help?