SSL is completely down from here

[Urs Echternacht]

Sorry, no web-pages are loaded over here, not on two different linux versions or two different windows variants. Still have to wait.

[speedimic]

as I said; REALLY strange.
I'll go to bed now and dream the solution...

gute nacht.

[Sutaru Tsureku]

SSL ?? 

I can't reach:
http://boinc.berkeley.edu
http://setiathome.berkeley.edu

 


The problem is posted at Berkeley?
[technical thread?]
Or maybe PM to the Berkeley-crew?

[speedimic]

I'm in office now and there is no simple way to reach SSL (Space Science Laboratory).

I'm behind a Linux proxy here, so that might be the problem... I'll go to another network segment to get around the proxy later and look if windows does the job...

[Sutaru Tsureku]

I can't reach Berkeley..
WinXP and IE.

I feel alone now.. 
 

There are some news because of the prob?
Berkeley-crew know because of the prob and/or some post/s in the forum?

[Richard Haselgrove]

I'm situated in the UK. I can contact Berkeley (both SETI and BOINC) with no problems: the SETI server page is showing everything running except the usual couple of splitters.

No, there's no problem with Berkeley. There clearly is some problem with the internet connection between Germany and Berkeley, but I doubt it is under the control of the SETI staff.

[Jason G]

Yeah it's been fine here in Australia too. It's really weird if these people can get up/downloads & pings through but no web pages ...

[Richard Haselgrove]

A possible clue: I have one machine (only) which is failing to download new SETI work from Berkeley. Uploads and scheduler contacts (and general browser work) are fine. The problem is only with Vista/v5.10.13: XP/v5.10.13, and XP/v6.2.18, are both fine.

I turned on debug options in cc_config.xml, and received:

09/10/2008 10:59:34||[file_xfer_debug] PERS_FILE_XFER::start_xfer(): URL: http://boinc2.ssl.berkeley.edu/sah/download_fanout/2b0/19au08aa.25729.25021.10.8.171
09/10/2008 10:59:34||[http_debug] [ID#2] info: About to connect() to boinc2.ssl.berkeley.edu port 80 (#0)
09/10/2008 10:59:34||[http_debug] [ID#2] info:   Trying 208.68.240.13...
09/10/2008 10:59:34||[http_debug] [ID#2] info: Connected to boinc2.ssl.berkeley.edu (208.68.240.13) port 80 (#0)
09/10/2008 10:59:34||[http_debug] [ID#2] Sent header to server: GET /sah/download_fanout/2b0/19au08aa.25729.25021.10.8.171 HTTP/1.1
User-Agent: BOINC client (windows_intelx86 5.10.13)
Host: boinc2.ssl.berkeley.edu
Accept: */*
Accept-Encoding: deflate, gzip
Content-Type: application/x-www-form-urlencoded
09/10/2008 10:59:34||[http_debug] [ID#2] Received header from server: HTTP/1.0 503 Service Unavailable
09/10/2008 10:59:34||[http_debug] [ID#2] Received header from server: Content-Type: text/html
09/10/2008 10:59:34||[http_debug] [ID#2] Received header from server: Content-Length: 53
09/10/2008 10:59:34||[http_xfer_debug] HTTP: wrote 53 bytes
09/10/2008 10:59:34||[http_debug] [ID#2] info: Expire cleared
09/10/2008 10:59:34||[http_debug] [ID#2] info: Closing connection #0
09/10/2008 10:59:35||[file_xfer_debug] FILE_XFER_SET::poll(): http op done; retval -184
09/10/2008 10:59:35||[file_xfer_debug] PERS_FILE_XFER::poll(): file transfer status -184

And then, while I was typing that.....

09/10/2008 11:07:49||[file_xfer_debug] PERS_FILE_XFER::start_xfer(): URL: http://boinc2.ssl.berkeley.edu/sah/download_fanout/1fa/19au08aa.25729.25021.10.8.189
09/10/2008 11:07:49||[http_debug] [ID#4] info: About to connect() to boinc2.ssl.berkeley.edu port 80 (#0)
09/10/2008 11:07:49||[http_debug] [ID#4] info:   Trying 208.68.240.18...
09/10/2008 11:07:49||[http_debug] [ID#4] info: Connected to boinc2.ssl.berkeley.edu (208.68.240.18) port 80 (#0)
09/10/2008 11:07:49||[http_debug] [ID#4] Sent header to server: GET /sah/download_fanout/1fa/19au08aa.25729.25021.10.8.189 HTTP/1.1
User-Agent: BOINC client (windows_intelx86 5.10.13)
Host: boinc2.ssl.berkeley.edu
Accept: */*
Accept-Encoding: deflate, gzip
Content-Type: application/x-www-form-urlencoded
09/10/2008 11:07:50||[http_debug] [ID#4] Received header from server: HTTP/1.1 200 OK
09/10/2008 11:07:50||[http_debug] [ID#4] Received header from server: Date: Thu, 09 Oct 2008 10:07:48 GMT
09/10/2008 11:07:50||[http_debug] [ID#4] Received header from server: Server: Apache/2.2.9 (Fedora)
09/10/2008 11:07:50||[http_debug] [ID#4] Received header from server: Last-Modified: Thu, 09 Oct 2008 06:01:09 GMT
09/10/2008 11:07:50||[http_debug] [ID#4] Received header from server: ETag: "11a9b789-5ba99-458cbc2581340"
09/10/2008 11:07:50||[http_debug] [ID#4] Received header from server: Accept-Ranges: bytes
09/10/2008 11:07:50||[http_debug] [ID#4] Received header from server: Content-Length: 375449
09/10/2008 11:07:50||[http_debug] [ID#4] Received header from server: Connection: close
09/10/2008 11:07:50||[http_debug] [ID#4] Received header from server: Content-Type: text/plain; charset=UTF-8

and a couple of them downloaded OK.

Do German ISPs use eclusively HTTP/1.0? Is it under anyone's direct control?

Edit - note the two different IP addresses:

208.68.240.13 is trying - and failing - to use HTTP/1.0 when the GET specifies HTTP/1.1
208.68.240.18 uses HTTP/1.1 as directed

Further edit:

I put 208.68.240.18   boinc2.ssl.berkeley.edu into a HOSTS file, flushed DNS, and restarted BOINC. All downloads completed at the first attempt.

[Jason G]

Hmmm, I think Urs mentioned that a web page request sits doing nothing rather than getting back a 503 service unavailable.  If they were getting that I'd suspect something 'closed for repair or maintenance'. Don't know about locale specific variants, but HTTP 1.0 & 1.1 are both application layer so only the client and server endpoints should be able to tell the difference...

 My next random guess is someone (Matt?) fiddling with QoS prioritisation somewhere, so some overloaded piece of equipment on a specific path might simply be dropping the requests...  Perhaps the offsite storage or similar could be via the same link...
If it's something weird like that then it should come good as soon as whatever congestion clears, but I woudln't count on it being anything like that situation.  Just a guess, which doesn't help matters.

[speedimic]

I just talked to a fellow admin located in Nürnberg.
He can get to setiathome.berkeley.edu without any problem.
The difference we figured out is that he is going via N-IX and I go via DE-CIX.
So the problem most likely is ISP dependent.

Edit:
just tried out what Richard tested.

I opened the two round robin IPs of boinc2.ssl.berkeley.edu with Firefox:

http://208.68.240.18/sah/download_fanout/2b0/ --> gives me the directory listing
http://208.68.240.13/sah/download_fanout/2b0/ --> The service is not available. Please try again later.

Further Edit:

It does not matter if I use HTTP 1.0 or 1.1.

[Jason G]

It'll be interesting to find out how the blockage is so specific. (if you ever find out that is)

[Richard Haselgrove]

I've sent the HTTP debug output in a PM to Matt and Eric. Hopefully they can figure it out when they get into the lab in 4 or 5 hours time.

[speedimic]

ok, did some more testing...

Richard's up/download problem has got nothing to do with our http problem.

boinc2.ssl.berkeley.edu is in a completely different network (208.68.something... (Hurricane Electric ??)) and works
setiathome.berkeley.edu is in 128.32.something... (Campus network) and doesn't work
fragment1.berkeley.edu (the Cricket grapher) is in 128.32.something... (Campus network) and doesn't work

that's why up/downloads work, but the website doesn't.

Taking that "different ISP- Different status" thing into account, I'd say it's some kind of peering problem here in Germany (especially Telekom / DE-CIX).

[Urs Echternacht]

If i use 208.68.240.18 port80 as a http proxy in Firefox on linux i can access setiathome.berkeley.edu. Do i use direct internet connection as usual no contact is made.

And instead of boinc.berkeley.edu i got the attached testpage with that as a proxy.

[attachment deleted by admin]

[speedimic]

If i use 208.68.240.18 port80 as a http proxy in Firefox on linux i can access setiathome.berkeley.edu. Do i use direct internet connection as usual no contact is made.

Already thought about that when I relized that boinc2.ssl.berkeley.edu is a proxy server but first, I'm alredy behind a proxy/firewall, and 2nd don't know if it's ok for berkeley if we use their proxy...