• Type: Bug
• Resolution: Answered
• Priority: Highest
• Fix Version/s: None
• Affects Version/s: 6.0.0
• Component/s: Webhooks
• Labels:

  • dob-not-suitable

[BSERV-12131] Webhooks intermittently stop working with Java 11 JRE

Michael Andreacchio made changes - 22/Jul/2021 2:47 AM

Remote Link

New: This issue links to "Page (Confluence)" [ 568517 ]

Tejaswini P made changes - 10/Dec/2020 8:59 AM

Remote Link

New: This issue links to "Page (Confluence)" [ 519732 ]

Ben Humphreys made changes - 20/Jul/2020 8:09 AM

Description

Original: {panel:title=Solution|borderStyle=solid|borderColor=#3c78b5|titleBGColor=#3c78b5|bgColor=#e7f4fa} This problem is caused by a bug in the Java runtime: _JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel_ The fix has been verified to be available in AdoptOpenJDK 11.0.8 - [https://github.com/AdoptOpenJDK/openjdk-jdk11u/commit/8d1b63a4db2c6348a97b3cf45bd4d2caa7cad6b5] {panel} Webhooks are likely impacted by the same problem as is described here: https://jira.atlassian.com/browse/BSERV-11796  In fixing BSERV-11796, all that was done was to disable TLSv1.3 in the HTTP Client, however that client instance is specific to the repository importer. The above is suspected to relate to the following JDK bug:  * JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel h3. Workaround 1) Run Bitbucket Server with Java 8; or  2) (Not yet verified) The issue is observed with Java 11 that uses TLS v1.3. It is likely that TLS v1.2 works fine and will not cause this issue. This can be done by setting the following JVM parameter: {{-Dhttps.protocols=TLSv1.2}} while starting bitbucket server up.

New: {panel:title=Solution|borderStyle=solid|borderColor=#3c78b5|titleBGColor=#3c78b5|bgColor=#e7f4fa} This problem is caused by a bug in the Java runtime: _JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel_ (Backported as [JDK-8241054|https://bugs.openjdk.java.net/browse/JDK-8241054]) The fix has been verified to be available in AdoptOpenJDK 11.0.8 - [https://github.com/AdoptOpenJDK/openjdk-jdk11u/commit/8d1b63a4db2c6348a97b3cf45bd4d2caa7cad6b5] {panel} Webhooks are likely impacted by the same problem as is described here: https://jira.atlassian.com/browse/BSERV-11796  In fixing BSERV-11796, all that was done was to disable TLSv1.3 in the HTTP Client, however that client instance is specific to the repository importer. The above is suspected to relate to the following JDK bug:  * JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel h3. Workaround 1) Run Bitbucket Server with Java 8; or  2) (Not yet verified) The issue is observed with Java 11 that uses TLS v1.3. It is likely that TLS v1.2 works fine and will not cause this issue. This can be done by setting the following JVM parameter: {{-Dhttps.protocols=TLSv1.2}} while starting bitbucket server up.

Ben Humphreys made changes - 20/Jul/2020 7:53 AM

Description

Original: {panel:title=Solution} This problem is caused by a bug in the Java runtime: _JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel_ The fix has been verified to be available in AdoptOpenJDK 11.0.8 - [https://github.com/AdoptOpenJDK/openjdk-jdk11u/commit/8d1b63a4db2c6348a97b3cf45bd4d2caa7cad6b5] {panel} Webhooks are likely impacted by the same problem as is described here: https://jira.atlassian.com/browse/BSERV-11796  In fixing BSERV-11796, all that was done was to disable TLSv1.3 in the HTTP Client, however that client instance is specific to the repository importer. The above is suspected to relate to the following JDK bug:  * JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel h3. Workaround 1) Run Bitbucket Server with Java 8; or  2) (Not yet verified) The issue is observed with Java 11 that uses TLS v1.3. It is likely that TLS v1.2 works fine and will not cause this issue. This can be done by setting the following JVM parameter: {{-Dhttps.protocols=TLSv1.2}} while starting bitbucket server up.

New: {panel:title=Solution|borderStyle=solid|borderColor=#3c78b5|titleBGColor=#3c78b5|bgColor=#e7f4fa} This problem is caused by a bug in the Java runtime: _JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel_ The fix has been verified to be available in AdoptOpenJDK 11.0.8 - [https://github.com/AdoptOpenJDK/openjdk-jdk11u/commit/8d1b63a4db2c6348a97b3cf45bd4d2caa7cad6b5] {panel} Webhooks are likely impacted by the same problem as is described here: https://jira.atlassian.com/browse/BSERV-11796  In fixing BSERV-11796, all that was done was to disable TLSv1.3 in the HTTP Client, however that client instance is specific to the repository importer. The above is suspected to relate to the following JDK bug:  * JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel h3. Workaround 1) Run Bitbucket Server with Java 8; or  2) (Not yet verified) The issue is observed with Java 11 that uses TLS v1.3. It is likely that TLS v1.2 works fine and will not cause this issue. This can be done by setting the following JVM parameter: {{-Dhttps.protocols=TLSv1.2}} while starting bitbucket server up.

Ben Humphreys made changes - 20/Jul/2020 7:50 AM

Resolution		New: Answered [ 9 ]
Status	Original: Short Term Backlog [ 12074 ]	New: Closed [ 6 ]

Ben Humphreys made changes - 20/Jul/2020 7:50 AM

Affects Version/s		New: 6.0.0 [ 82493 ]
Affects Version/s	Original: 6.7.1 [ 89891 ]

Ben Humphreys made changes - 20/Jul/2020 7:50 AM

Description

Original: Webhooks are likely impacted by the same problem as is described here: https://jira.atlassian.com/browse/BSERV-11796  In fixing BSERV-11796, all that was done was to disable TLSv1.3 in the HTTP Client, however that client instance is specific to the repository importer. The above is suspected to relate to the following JDK bug:  * JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel h3. Workaround 1) Run Bitbucket Server with Java 8; or  2) (Not yet verified) The issue is observed with Java 11 that uses TLS v1.3. It is likely that TLS v1.2 works fine and will not cause this issue. This can be done by setting the following JVM parameter: {{-Dhttps.protocols=TLSv1.2}} while starting bitbucket server up.

New: {panel:title=Solution} This problem is caused by a bug in the Java runtime: _JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel_ The fix has been verified to be available in AdoptOpenJDK 11.0.8 - [https://github.com/AdoptOpenJDK/openjdk-jdk11u/commit/8d1b63a4db2c6348a97b3cf45bd4d2caa7cad6b5] {panel} Webhooks are likely impacted by the same problem as is described here: https://jira.atlassian.com/browse/BSERV-11796  In fixing BSERV-11796, all that was done was to disable TLSv1.3 in the HTTP Client, however that client instance is specific to the repository importer. The above is suspected to relate to the following JDK bug:  * JDK-8214418 HttpClient falls in running with 100% cpu usage after an error signalled on channel h3. Workaround 1) Run Bitbucket Server with Java 8; or  2) (Not yet verified) The issue is observed with Java 11 that uses TLS v1.3. It is likely that TLS v1.2 works fine and will not cause this issue. This can be done by setting the following JVM parameter: {{-Dhttps.protocols=TLSv1.2}} while starting bitbucket server up.

SET Analytics Bot made changes - 18/Jul/2020 12:43 AM

UIS

Original: 895

New: 831

SET Analytics Bot made changes - 12/Jul/2020 1:35 AM

UIS

Original: 772

New: 895

Bugfix Automation Bot made changes - 10/Jul/2020 4:30 AM

Support reference count

Original: 23

New: 24

Load more older history items