Différences

Ci-dessous, les différences entre deux révisions de la page.

--- raspberry:usb_issue [07-06-2017 21:54]
edmc73 créée
+++ raspberry:usb_issue [16-06-2017 00:03] (Version actuelle)
edmc73
@@ Ligne 111: / Ligne 111: @@
 Je verrai par la suite si ça résout mon problème....
+Cette solution semble marcher :)
+Maintenant je suis confronté à un autre problème
+<code>
+[32398.188252] INFO: task kworker/u2:1:13577 blocked for more than 120 seconds.
+[32398.188288]       Not tainted 4.1.19+ #858
+[32398.188301] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
+[32398.188315] kworker/u2:1    D c0534afc     0 13577      2 0x00000000
+[32398.189053] Workqueue: btrfs-endio-write btrfs_endio_write_helper [btrfs]
+[32398.189174] [<c0534afc>] (__schedule) from [<c0534e44>] (schedule+0x4c/0xa4)
+[32398.189213] [<c0534e44>] (schedule) from [<c0537224>] (schedule_timeout+0x180/0x1f8)
+[32398.189241] [<c0537224>] (schedule_timeout) from [<c0535864>] (wait_for_common+0x9c/0x150)
+[32398.189268] [<c0535864>] (wait_for_common) from [<c0535938>] (wait_for_completion+0x20/0x24)
+[32398.189698] [<c0535938>] (wait_for_completion) from [<bf05a360>] (btrfs_async_run_delayed_refs+0x100/0x134 [btrfs])
+[32398.190447] [<bf05a360>] (btrfs_async_run_delayed_refs [btrfs]) from [<bf079bfc>] (__btrfs_end_transaction+0x248/0x368 [btrfs])
+[32398.191306] [<bf079bfc>] (__btrfs_end_transaction [btrfs]) from [<bf079d38>] (btrfs_end_transaction+0x1c/0x20 [btrfs])
+[32398.192167] [<bf079d38>] (btrfs_end_transaction [btrfs]) from [<bf083a6c>] (btrfs_finish_ordered_io+0x174/0x67c [btrfs])
+[32398.193016] [<bf083a6c>] (btrfs_finish_ordered_io [btrfs]) from [<bf0842a4>] (finish_ordered_fn+0x1c/0x20 [btrfs])
+[32398.193938] [<bf0842a4>] (finish_ordered_fn [btrfs]) from [<bf0b2634>] (normal_work_helper+0x8c/0x358 [btrfs])
+[32398.194936] [<bf0b2634>] (normal_work_helper [btrfs]) from [<bf0b2c18>] (btrfs_endio_write_helper+0x1c/0x20 [btrfs])
+[32398.195510] [<bf0b2c18>] (btrfs_endio_write_helper [btrfs]) from [<c0038e7c>] (process_one_work+0x11c/0x394)
+[32398.195551] [<c0038e7c>] (process_one_work) from [<c0039134>] (worker_thread+0x40/0x4d4)
+[32398.195585] [<c0039134>] (worker_thread) from [<c003e5dc>] (kthread+0xdc/0xf8)
+[32398.195623] [<c003e5dc>] (kthread) from [<c000f858>] (ret_from_fork+0x14/0x3c)
+[32398.195646] INFO: task kworker/u2:3:13579 blocked for more than 120 seconds.
+[32398.195662]       Not tainted 4.1.19+ #858
+[32398.195673] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
+[32398.195687] kworker/u2:3    D c0534afc     0 13579      2 0x00000000
+[32398.196208] Workqueue: btrfs-endio-write btrfs_endio_write_helper [btrfs]
+</code>
+Apparemment, pendant mon backup, il y a trop d'IO disk et ça fini par crasher (pour faire simple)
+Une très bonne doc explique tout ça avec plusieurs solutions => https://lonesysadmin.net/2013/12/22/better-linux-disk-caching-performance-vm-dirty_ratio/
+Je vais tester ceci => https://www.blackmoreops.com/2014/09/22/linux-kernel-panic-issue-fix-hung_task_timeout_secs-blocked-120-seconds-problem/ qui permet, si j'ai bien compris, de réduire la mise en cache des fichiers disques et d'écrire plus souvent sur le disque.
+----
+Le problème semble encore là et la lenteur semble être bien présente
+<code>
+# hdparm -t -T /dev/sda
+/dev/sda:
+ Timing cached reads:     8 MB in  2.01 seconds =   3.99 MB/sec
+ Timing buffered disk reads:   4 MB in  3.82 seconds =   1.05 MB/sec
+</code>
+d'ailleurs un test d'écriture avec dd à planté..
+  # dd if=/dev/zero of=/media/memup/test.tmp bs=1M count=1024 conv=fdatasync
+<code>
+INFO: task kworker/u2:3:24564 blocked for more than 120 seconds.
+      Not tainted 4.1.19+ #858
+"echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
+kworker/u2:3    D c0534afc     0 24564      2 0x00000000
+Workqueue: writeback bdi_writeback_workfn (flush-btrfs-3)
+[<c0534afc>] (__schedule) from [<c0534e44>] (schedule+0x4c/0xa4)
+[<c0534e44>] (schedule) from [<bf05458c>] (wait_block_group_cache_progress+0xb8/0xf0 [btrfs])
+[<bf05458c>] (wait_block_group_cache_progress [btrfs]) from [<bf060b78>] (find_free_extent+0x4c4/0xec8 [btrfs])
+[<bf060b78>] (find_free_extent [btrfs]) from [<bf061730>] (btrfs_reserve_extent+0x78/0x204 [btrfs])
+[<bf061730>] (btrfs_reserve_extent [btrfs]) from [<bf080b54>] (cow_file_range+0x1d4/0x810 [btrfs])
+[<bf080b54>] (cow_file_range [btrfs]) from [<bf082120>] (run_delalloc_range+0x38c/0x45c [btrfs])
+[<bf082120>] (run_delalloc_range [btrfs]) from [<bf09d050>] (writepage_delalloc+0x174/0x230 [btrfs])
+[<bf09d050>] (writepage_delalloc [btrfs]) from [<bf09e50c>] (__extent_writepage+0xe4/0x2f0 [btrfs])
+[<bf09e50c>] (__extent_writepage [btrfs]) from [<bf0a0a4c>] (extent_writepages+0x2c4/0x3bc [btrfs])
+[<bf0a0a4c>] (extent_writepages [btrfs]) from [<bf07ccf4>] (btrfs_writepages+0x30/0x38 [btrfs])
+[<bf07ccf4>] (btrfs_writepages [btrfs]) from [<c00e0e50>] (do_writepages+0x34/0x48)
+[<c00e0e50>] (do_writepages) from [<c014c060>] (__writeback_single_inode+0x40/0x2ac)
+[<c014c060>] (__writeback_single_inode) from [<c014c44c>] (writeback_sb_inodes+0x180/0x390)
+[<c014c44c>] (writeback_sb_inodes) from [<c014c6e8>] (__writeback_inodes_wb+0x8c/0xd0)
+[<c014c6e8>] (__writeback_inodes_wb) from [<c014c90c>] (wb_writeback+0x1e0/0x27c)
+[<c014c90c>] (wb_writeback) from [<c014d030>] (bdi_writeback_workfn+0x1d0/0x474)
+[<c014d030>] (bdi_writeback_workfn) from [<c0038e7c>] (process_one_work+0x11c/0x394)
+[<c0038e7c>] (process_one_work) from [<c0039134>] (worker_thread+0x40/0x4d4)
+[<c0039134>] (worker_thread) from [<c003e5dc>] (kthread+0xdc/0xf8)
+[<c003e5dc>] (kthread) from [<c000f858>] (ret_from_fork+0x14/0x3c)
+</code>
+----
+Conclusion, en fait le problème venait d'une défaillance du disque.
+Le problème était visible dans le rapport SMART. Une usure prématuré rendant les performances du disques appauvri.
+Le disque avait les mêmes symptôme branché sur un PC, la génération d'un fichier avec dd échouait. Suite à ça, j'ai racheté un nouveau disque et fait l'erreur de faire un clonage de disque.
+Pour une raison que j'ignore, les mêmes symptomes étaient présent sur la nouvelle partition cloné, impossible de faire un dd.
+Finalement, création d'une partition toute fraiche et copie des fichiers via rsync, maintenant tout fonctionne comme avant, voir mieux :)

Outils pour utilisateurs

Outils du site

Différences

Outils de la page