[
https://issues.apache.org/jira/browse/JAMES-3891?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17693113#comment-17693113
]
Karsten Otto commented on JAMES-3891:
-------------------------------------
> As a fix I propose to discard emails not backed by a blob.
Does this mean we actually *lost* some mails during restart? That would be
bad...
> ActiveMQ: restarting james can block mail queue
> -----------------------------------------------
>
> Key: JAMES-3891
> URL: https://issues.apache.org/jira/browse/JAMES-3891
> Project: James Server
> Issue Type: Improvement
> Components: Queue
> Affects Versions: master, 3.7.3
> Reporter: Benoit Tellier
> Priority: Major
> Fix For: 3.8.0, 3.7.4
>
> Time Spent: 10m
> Remaining Estimate: 0h
>
> Given a mail queue currently containing emails and being consumed, When I
> restart James, Then after the restart the email consumptions sometime do not
> resume.
> The following log can be seen:
> {code:java}
> james1 | java.io.FileNotFoundException:
> /root/var/store/activemq/blob-transfer/10/ID_james1.local-33677-1677150190846-5_1_1_1_1037
> (No such file or directory)
> james1 | at java.base/java.io.FileInputStream.open0(Native
> Method)
> james1 | at java.base/java.io.FileInputStream.open(Unknown
> Source)
> james1 | at java.base/java.io.FileInputStream.<init>(Unknown
> Source)
> james1 | at
> org.apache.james.queue.activemq.FileSystemBlobStrategy.getInputStream(FileSystemBlobStrategy.java:101)
> james1 | at
> org.apache.activemq.blob.BlobDownloader.getInputStream(BlobDownloader.java:38)
> james1 | at
> org.apache.activemq.command.ActiveMQBlobMessage.getInputStream(ActiveMQBlobMessage.java:132)
> james1 | at
> org.apache.james.queue.activemq.MimeMessageBlobMessageSource.getInputStream(MimeMessageBlobMessageSource.java:45)
> james1 | at
> org.apache.james.server.core.MimeMessageWrapper.writeTo(MimeMessageWrapper.java:307)
> james1 | at
> org.apache.james.server.core.MimeMessageWrapper.writeTo(MimeMessageWrapper.java:296)
> james1 | at
> org.apache.james.server.core.MimeMessageWrapper.writeTo(MimeMessageWrapper.java:292)
> james1 | at
> org.apache.james.server.core.MimeMessageWrapper.writeTo(MimeMessageWrapper.java:276)
> james1 | at
> org.apache.james.server.core.MimeMessageWrapper.<init>(MimeMessageWrapper.java:158)
> james1 | ... 19 common frames omitted
> james1 | Wrapped by: javax.mail.MessagingException: IOException
> while copying message
> james1 | at
> org.apache.james.server.core.MimeMessageWrapper.<init>(MimeMessageWrapper.java:165)
> james1 | at
> org.apache.james.server.core.MailImpl.setMessage(MailImpl.java:505)
> james1 | at
> com.github.fge.lambdas.consumers.ConsumerChainer.lambda$sneakyThrow$9(ConsumerChainer.java:73)
> james1 | at java.base/java.util.Optional.ifPresent(Unknown
> Source)
> james1 | at
> org.apache.james.server.core.MailImpl$Builder.build(MailImpl.java:279)
> james1 | at
> org.apache.james.queue.jms.JMSCacheableMailQueue.createMail(JMSCacheableMailQueue.java:391)
> james1 | at
> org.apache.james.queue.activemq.ActiveMQCacheableMailQueue.createMailQueueItem(ActiveMQCacheableMailQueue.java:230)
> james1 | at
> org.apache.james.queue.jms.JMSCacheableMailQueue.deQueueOneItem(JMSCacheableMailQueue.java:249)
> james1 | ... 12 common frames omitted
> james1 | Wrapped by:
> org.apache.james.queue.api.MailQueue$MailQueueException: Unable to dequeue
> next message
> james1 | at
> org.apache.james.queue.jms.JMSCacheableMailQueue.deQueueOneItem(JMSCacheableMailQueue.java:260)
> james1 | at
> reactor.core.publisher.MonoDefer.subscribe(MonoDefer.java:44)
> james1 | at
> reactor.core.publisher.FluxRepeatPredicate$RepeatPredicateSubscriber.resubscribe(FluxRepeatPredicate.java:120)
> james1 | at
> reactor.core.publisher.MonoRepeatPredicate.subscribeOrReturn(MonoRepeatPredicate.java:48)
> james1 | at
> reactor.core.publisher.InternalFluxOperator.subscribe(InternalFluxOperator.java:55)
> james1 | at
> reactor.core.publisher.FluxSubscribeOn$SubscribeOnSubscriber.run(FluxSubscribeOn.java:194)
> james1 | at
> reactor.core.scheduler.WorkerTask.call(WorkerTask.java:84)
> james1 | at
> reactor.core.scheduler.WorkerTask.call(WorkerTask.java:37)
> james1 | at
> java.base/java.util.concurrent.FutureTask.run(Unknown Source)
> james1 | at
> java.base/java.util.concurrent.ScheduledThreadPoolExecutor$ScheduledFutureTask.run(Unknown
> Source)
> james1 | at
> java.base/java.util.concurrent.ThreadPoolExecutor.runWorker(Unknown Source)
> james1 | at
> java.base/java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
> james1 | at java.base/java.lang.Thread.run(Unknown Source)
> james1 | Wrapped by:
> reactor.core.Exceptions$ErrorCallbackNotImplemented:
> org.apache.james.queue.api.MailQueue$MailQueueException: Unable to dequeue
> next message;
> james1 | nested exception is:
> james1 | javax.mail.MessagingException: IOException while
> copying message;
> james1 | nested exception is:
> james1 | java.io.FileNotFoundException:
> /root/var/store/activemq/blob-transfer/10/ID_james1.local-33677-1677150190846-5_1_1_1_1037
> (No such file or directory)
> {code}
> No processing takes place and remaining emails are stuck into the mail queue.
> The issue is that blobs are handle via file so their deletion is not linked
> to the JMS trasaction: the JMS transaction can be rolled back for a deleted
> blob, thus leading to messages without blob.
> Steps to reproduce:
> - Send 5000 mails to james of 500 bytes each
> {code:java}
> smtp-source -A -C100 -r 1 -l 500 -m 5000 -s 20 -d -c -f a...@other.com -M
> other.com -t bob@localhost 172.24.0.4:25
> {code}
> - Once all mails are queued, restart james several time during processing
> - Increasing the count of spooler workers helps.
> As a fix I propose to discard emails not backed by a blob.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
---------------------------------------------------------------------
To unsubscribe, e-mail: server-dev-unsubscr...@james.apache.org
For additional commands, e-mail: server-dev-h...@james.apache.org
