https://bugzilla.kernel.org/show_bug.cgi?id=118081 --- Comment #1 from Mike Christie <mchristi@xxxxxxxxxx> --- On 05/11/2016 10:34 PM, bugzilla-daemon@xxxxxxxxxxxxxxxxxxx wrote: > https://bugzilla.kernel.org/show_bug.cgi?id=118081 > > Bug ID: 118081 > Summary: open-iscsi Ping timeout erro > Product: SCSI Drivers > Version: 2.5 > Kernel Version: 4.4.7 > Hardware: All > OS: Linux > Tree: Mainline > Status: NEW > Severity: normal > Priority: P1 > Component: Other > Assignee: scsi_drivers-other@xxxxxxxxxxxxxxxxxxxx > Reporter: liuzhengyuang521@xxxxxxxxx > Regression: No > > Hi everyone: > I create a target using fileio as the backend storage on ARM64 server. The > initiator reported some errors showed bellow while perform iozone test. > > [178444.145679] connection14:0: ping timeout of 5 secs expired, recv timeout > 5, last rx 4339462894, last ping 4339464146, now 4339465400 > [178444.145706] connection14:0: detected conn error (1011) > [178469.674313] connection14:0: detected conn error (1020) > [178504.420979] connection14:0: ping timeout of 5 secs expired, recv timeout > 5, last rx 4339477953, last ping 4339479204, now 4339480456 > [178504.421001] connection14:0: detected conn error (1011) > [178532.064262] connection14:0: detected conn error (1020) > [178564.584087] connection14:0: ping timeout of 5 secs expired, recv timeout > 5, last rx 4339492980, last ping 4339494232, now 4339495484 > .............................. > > I try to trace the function call of target iscsi. Then, I found the receiving > thread of target iscsi blocked at fd_execute_sync_cache -> vfs_fsync_range. > Further, vfs_fsync_range may takes more than 10 seconds to return,while > initiator Ping timeout would happened after 5 seconds. vfs_fsync_range was > call with the form vfs_fsync_range(fd_dev->fd_file, 0, LLONG_MAX, 1) every > times which means sync all device cache. > So, is this a bug? > How does Initiator send sync_cache scsi command? > Does it need to sync all device cache at once? > Any reply would be thankful. > The upper layers like the FS or application determine when to send a sync cache. They send down a request and the iscsi layer just sends it to the target. You are using LIO right? It looks like we end up syncing the entire device sometimes. I think for iscsi pings/Nops that have the immediate bit set, the target would want to reply to them right away. They should not be getting stuck behind these type of commands. Nick, what do you think? -- You are receiving this mail because: You are watching the assignee of the bug. -- To unsubscribe from this list: send the line "unsubscribe linux-scsi" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
