Has anyone seen a problem with CosignModule for IIS where users sometimes, but not always, receive a 503 "Service unavailable" message when they first try to access a cosign-protected resource? (Specifically, this error sometimes occurs when the user is redirected from the central weblogin servers to /cosign/valid on the server that is running CosignModule)
We're running CosignModule 3.1.0 under Windows Server 2008R2 SP1 (IIS 7.5). Here are the DebugView logs for a failed request: 00001464 7.48706007 [19560] [CosignModule] CosignModule created. 00001465 7.48711443 [19560] [CosignModule] OnAuthenticateRequest() Thread id = 40172l 00001466 7.48714542 [19560] [CosignModule] New GetConfig()uration logics! 00001467 7.48717213 [19560] [CosignModule] Metapath = MACHINE/WEBROOT/APPHOST/DEFAULT WEB SITE/COSIGN/VALID 00001468 7.48728895 [19560] [CosignModule] unprotected url 00001469 7.48735619 [19560] [CosignModule] execreqhandler: qs = cosign-webapps.lsa.umich.edu=6Uf2+BIoBjEQ7c34Qj1zIUK2myNWUW3EEqJ7WjnKu+9dLXKnQ97ADa+N7+RyoyWIQ968B-ITkXnkxU2lPKSvvi-yLjVrwL4VQ41Evq7KOHLqsdCRHsuGCHdtHqjL&https://webapps.lsa.umich.edu/advappts 00001470 7.48739100 [19560] [CosignModule] {*********************GetValidationConfig*********************} 00001471 7.48741770 [19560] [CosignModule] Metapath = MACHINE/WEBROOT/APPHOST/DEFAULT WEB SITE/COSIGN/VALID 00001472 7.48753500 [19560] [CosignModule] NumFactors = 0 00001473 7.48756695 [19560] [CosignModule] {*********************GetValidationConfig Done****************} 00001474 7.48759222 [19560] [CosignModule] Regex is https?:\/\/.*\.umich\.edu(/.*)? 00001475 7.48763275 [19560] [CosignModule] CHECK'ing cookie 00001476 7.48765802 [19560] [CosignModule] CHECKing cookie, waiting for mutex. 00001477 7.48768282 [19560] [CosignModule] Obtained the mutex. 00001478 7.48770666 [19560] [CosignModule] connections.size() = 6 00001479 7.48773193 [19560] [CosignModule] CheckCookie iter 0 00001480 7.48775911 [19560] [CosignModule]>> CHECK cosign-webapps.lsa.umich.edu=6Uf2+BIoBjEQ7c34Qj1zIUK2myNWUW3EEqJ7WjnKu+9dLXKnQ97ADa+N7+RyoyWIQ968B-ITkXnkxU2lPKSvvi-yLjVrwL4VQ41Evq7KOHLqsdCRHsuGCHdtHqjL 00001481 7.48786736 [19560] [CosignModule] recv()'ing bytes at offset 0 00001482 7.48786736 [19560] readBufferSize = 16425 00001483 7.48786736 [19560] readBufferSize - readBufferOffset = 16425 00001484 7.50066996 [19560] [CosignModule] Received 106 encrypted bytes 00001485 7.50070763 [19560] [CosignModule] ss is SEC_E_OK 00001486 7.50073433 [19560] [CosignModule] Found 0 bytes of SECBUFFER_DATA at 1 00001487 7.50075912 [19560] [CosignModule] Found 69 bytes in SECBUFFER_EXTRA[ 3 ] 00001488 7.50078440 [19560] [CosignModule] Putting 69 bytes from SECBUFFER_EXTRA[ 3 ] into extraBuffer 00001489 7.50080824 [19560] [CosignModule] Decrypted 0 bytes as: 00001490 7.50083160 [19560] [CosignModule] cbBuffer is 0 00001491 7.50085545 [19560] [CosignModule] Found extra data! 00001492 7.50087929 [19560] [CosignModule] secureRead looping(0) 00001493 7.50090885 [19560] [CosignModule] ss is SEC_E_OK 00001494 7.50093365 [19560] [CosignModule] Found 30 bytes of SECBUFFER_DATA at 1 00001495 7.50095749 [19560] [CosignModule] Decrypted 30 bytes as: 533 CHECK: cookie not in db! 00001496 7.50098133 [19560] [CosignModule] No extra data found! 00001497 7.50100517 [19560] [CosignModule]<< 533 CHECK: cookie not in db! 00001498 7.50102854 [19560] [CosignModule] Trying a different server: 533 CHECK: cookie not in db! 00001499 7.50105381 [19560] [CosignModule] CheckCookie iter 1 00001500 7.50108051 [19560] [CosignModule]>> CHECK cosign-webapps.lsa.umich.edu=6Uf2+BIoBjEQ7c34Qj1zIUK2myNWUW3EEqJ7WjnKu+9dLXKnQ97ADa+N7+RyoyWIQ968B-ITkXnkxU2lPKSvvi-yLjVrwL4VQ41Evq7KOHLqsdCRHsuGCHdtHqjL 00001501 7.50112438 [19560] [CosignModule] recv()'ing bytes at offset 0 00001502 7.50112438 [19560] readBufferSize = 16425 00001503 7.50112438 [19560] readBufferSize - readBufferOffset = 16425 00001504 7.50115442 [19560] [CosignModule] Socket error on recv(): 10053 00001505 7.50117874 [19560] [CosignModule] Error reading data(4) from socket 1 00001506 7.50120258 [19560] [CosignModule] CheckCookie iter 2 00001507 7.50122643 [19560] [CosignModule]>> STARTTLS 2 00001508 7.50126743 [19560] [CosignModule] Error reading data from socket 2 00001509 7.50129271 [19560] [CosignModule] CheckCookie iter 3 00001510 7.50131559 [19560] [CosignModule]>> STARTTLS 2 00001511 7.50135183 [19560] [CosignModule] Error reading data from socket 3 00001512 7.50137663 [19560] [CosignModule] CheckCookie iter 4 00001513 7.50139999 [19560] [CosignModule]>> STARTTLS 2 00001514 7.50143576 [19560] [CosignModule] Error reading data from socket 4 00001515 7.50146008 [19560] [CosignModule] CheckCookie iter 5 00001516 7.50148392 [19560] [CosignModule]>> STARTTLS 2 00001517 7.50152016 [19560] [CosignModule] Error reading data from socket 5 00001518 7.50154543 [19560] [CosignModule] Released the mutex. 00001519 7.50156927 [19560] [CosignModule] Service cookie invalid: all weblogin servers returned RETRY. 00001520 7.50321245 [19560] [CosignModule] CosignModule destructed. Here is the code that is generating the two different "Error reading data" messages above: http://cosign.git.sourceforge.net/git/gitweb.cgi?p=cosign/cosignmodule;a=blob;f=ConnectionList.cpp;h=32f9aad027e0f1597adf27d5386873261d16f7cb;hb=HEAD#l142 Several things strike me as being strange: - So far, we have only observed the failure on the initial URL validation. If the user re-loads the page, the validation succeeds and the user is able to use the web application normally without problem. - ConnectionList::CheckCookie is using a pool of already-open connections but is experiencing read errors. The first read error ("data(4)", above) is on a connection on which TLS has already been started -- CosignModule sends a CHECK command, and then is unable to read the response. It then tries four other connections, none of which have had TLS started, sends the STARTTLS command, and then is unable to read the responses. I've checked the DebugView logs for four separate 503 errors and this patten holds exactly for all of them (failure to read CHECK response on the first connection, followed by failure to read STARTTLS responses on four other connections). Any ideas or assistance would be appreciated, but if anyone else has seen this issue, too, then that's something we'd definitely like to know. -- Mark Montague [email protected] ------------------------------------------------------------------------------ Live Security Virtual Conference Exclusive live event will cover all the ways today's security and threat landscape has changed and how IT managers can respond. Discussions will include endpoint security, mobile security and the latest in malware threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/ _______________________________________________ Cosign-discuss mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/cosign-discuss
