+++ This bug was initially created as a clone of Bug #901145 +++ Complexity: High project_key: JBEWS Jira * [JBPAPP-10319] mod_cluster returns "Bad Gateway" HTTP ErrorCode 502 with https is likely to affect EWS 2.0 as well. h3. Issue confirmed Issue reproduced with *EWS 2.0.0 CR4*. See the attached [^JBPAPP-9493-reproduced-with-EWS2.0.0.CR4.zip] --- Additional comment from Michal Babacek on 2012-10-31 19:30:36 EDT --- Link: Added: This issue Cloned from JBPAPP-10319 --- Additional comment from Michal Babacek on 2012-10-31 19:30:36 EDT --- Link: Added: This issue is related to JBPAPP-10029 --- Additional comment from Michal Babacek on 2012-10-31 19:30:37 EDT --- Link: Added: This issue is related to JBPAPP-9493 --- Additional comment from Michal Babacek on 2012-10-31 19:33:33 EDT --- Steps to Reproduce: Removed: h3. How to reproduce # Configure AS7 with HTTPS connector as in [^JBPAPP-9493-reproduced-with-m_c-1.2.3.Final.zip] # Configure Httpd + Mod_cluster with SSL as in [^JBPAPP-9493-reproduced-with-m_c-1.2.3.Final.zip] # Start httpd # Start AS7 # Send a request, e.g. like: {noformat} curl https://localhost:8888/SessionTest/session --cert /home/mbabacek/EAP6/Client/client.crt --key /home/mbabacek/EAP6/Client/client.key --cacert /home/mbabacek/EAP6/Client/myca.crt --insecure -c originally_empty_cookiefile.txt -b originally_empty_cookiefile.txt 2> /dev/null {noformat} # Optionally, wait ~10 minutes. (!) You might start to having errors even without this request free wait period. # Sned another request & get an error # Eventually, after *STATUS* message is received, you will get correct HTTP 200 again. (i) *Note:* As soon as STATUS message is received, it picks up again... Workaround Description: Removed: Turning off keepAlive settings, see [Aaron's comment|https://issues.jboss.org/browse/JBPAPP-9493?focusedCommentId=12709722&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-12709722] --- Additional comment from lfuka on 2012-11-05 07:35:09 EST --- Link: Added: This issue relates to JBPAPP-9551 --- Additional comment from Michal Babacek on 2012-11-05 08:54:55 EST --- Release Notes Text: Added: Issue causes mod_cluster to return an HTTP error 502 (Bad Gateway) after a period of inactivity in case only HTTPS connector is enabled for balancer-worker communication. Httpd tries to re-use a connection that have been closed already. SSL handshake fails and HTTP error is triggered. Known workaround: Either allow AJP connector on workers or: Set maxKeepAliveRequests="1" in server.xml, Set smax to 0 in mod_cluster worker configuration, SetEnv proxy-nokeepalive 1 and SetEnv proxy-initial-not-pooled 1 in your httpd configuration. --- Additional comment from Michal Babacek on 2012-11-05 10:15:18 EST --- Attachment: Added: JBPAPP-9493-reproduced-with-EWS2.0.0.CR4.zip --- Additional comment from Misha H. Ali on 2012-11-05 10:23:39 EST --- Release Notes Docs Status: Added: Documented as Known Issue Writer: Added: mhusnain Release Notes Text: Removed: Issue causes mod_cluster to return an HTTP error 502 (Bad Gateway) after a period of inactivity in case only HTTPS connector is enabled for balancer-worker communication. Httpd tries to re-use a connection that have been closed already. SSL handshake fails and HTTP error is triggered. Known workaround: Either allow AJP connector on workers or: Set maxKeepAliveRequests="1" in server.xml, Set smax to 0 in mod_cluster worker configuration, SetEnv proxy-nokeepalive 1 and SetEnv proxy-initial-not-pooled 1 in your httpd configuration. Added: mod_cluster returns a HTTP error 502 (Bad Gateway) after a period of inactivity if only HTTPS connector is enabled for balancer-worker communication. Httpd tries to re-use a connection that have been closed already. SSL handshake fails and HTTP error is triggered. As a workaround, either allow AJP connector on all workers or set maxKeepAliveRequests="1" in server.xml, Set smax to 0 in mod_cluster worker configuration, SetEnv proxy-nokeepalive 1 and SetEnv proxy-initial-not-pooled 1 in your httpd configuration. --- Additional comment from Michal Babacek on 2012-11-05 10:55:09 EST --- Workaround in *Release Notes* by [~aogburn] in [comment|https://issues.jboss.org/browse/JBPAPP-9493?focusedCommentId=12709722&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-12709722]. --- Additional comment from Michal Babacek on 2012-11-07 09:59:12 EST --- Follow the most up-to-date feed on [JBPAPP-10029|https://issues.jboss.org/browse/JBPAPP-10029?focusedCommentId=12732465&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-12732465] ... --- Additional comment from Jiri Skrabal on 2012-11-13 10:56:38 EST --- Release Notes Docs Status: Removed: Documented as Known Issue Writer: Removed: mhusnain Release Notes Text: Removed: mod_cluster returns a HTTP error 502 (Bad Gateway) after a period of inactivity if only HTTPS connector is enabled for balancer-worker communication. Httpd tries to re-use a connection that have been closed already. SSL handshake fails and HTTP error is triggered. As a workaround, either allow AJP connector on all workers or set maxKeepAliveRequests="1" in server.xml, Set smax to 0 in mod_cluster worker configuration, SetEnv proxy-nokeepalive 1 and SetEnv proxy-initial-not-pooled 1 in your httpd configuration. Docs QE Status: Removed: NEW --- Additional comment from Adam Kovari on 2012-11-14 03:35:55 EST --- Link: Added: This issue is related to JBPAPP-10409 --- Additional comment from Jean-frederic Clere on 2013-04-23 08:32:28 EDT --- It needs either the patch from: https://github.com/modcluster/mod_cluster/commit/855cdda451eb561abe10463133f36360d5a302fe or: a release 1.2.4.Final from mod_cluster.
We are going to need this fix in EWS 2.0.1 and in order to sync the version between EAP and EWS, we need this upgrade in EAP as well, hence this BZ.
mod_cluster-1.2.4-1.Final_redhat_1.ep6.el6 mod_cluster-native-1.2.4-1.Final.redhat_1.ep6.el6 mod_cluster-1.2.4-1.Final_redhat_1.ep6.el5 mod_cluster-native-1.2.4-1.Final.redhat_1.ep6.el5 built in Brew, jboss-component-management updated. Will respin webserverconnector zips after I have mod_jk updated as well.
jboss-eap-native-webserver-connectors-6.1.0-3.ep6.el6 jboss-eap-native-webserver-connectors-6.1.0-3.ep6.el5 built in Brew, jboss-eap-native-6.1.0-5.ep6.el6 also contains latest sources for mod_cluster, will be part of ER6.
Works like a charm with EAP 6.1.0.ER6. Verified.
Details inside vpn: https://post-office.corp.redhat.com/mailman/private/eap-pm-list/2013-July/msg00105.html