Scribd Logo
Upload

Welcome to Scribd!

  • 98 views

    Error-Last Erred, Retry IO Without B - FAILFAST

    Uploaded by

    vijayen123
    The document describes an issue where Solaris Volume Manager (SVM) metadevices were unexpectedly going into maintenance mode without disk errors. This was caused by the md_maxphys parameter being set incorrectly in the /etc/system file to a non-integer multiple of the sector size, resulting in I/O errors. The solution was to change md_maxphys to match the maxphys parameter, reboot and recreate the metadevices.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd

    Error-Last Erred, Retry IO Without B - FAILFAST

    Uploaded by

    vijayen123
    98 views2 pages
    The document describes an issue where Solaris Volume Manager (SVM) metadevices were unexpectedly going into maintenance mode without disk errors. This was caused by the md_maxphys parameter being set incorrectly in the /etc/system file to a non-integer multiple of the sector size, resulting in I/O errors. The solution was to change md_maxphys to match the maxphys parameter, reboot and recreate the metadevices.

    Original Description:

    Error-Last Erred, Retry IO Without B_FAILFAST

    Original Title

    Error-Last Erred, Retry IO Without B_FAILFAST

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF, TXT or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    The document describes an issue where Solaris Volume Manager (SVM) metadevices were unexpectedly going into maintenance mode without disk errors. This was caused by the md_maxphys parameter being set incorrectly in the /etc/system file to a non-integer multiple of the sector size, resulting in I/O errors. The solution was to change md_maxphys to match the maxphys parameter, reboot and recreate the metadevices.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    98 views2 pages

    Error-Last Erred, Retry IO Without B - FAILFAST

    Uploaded by

    vijayen123
    The document describes an issue where Solaris Volume Manager (SVM) metadevices were unexpectedly going into maintenance mode without disk errors. This was caused by the md_maxphys parameter being set incorrectly in the /etc/system file to a non-integer multiple of the sector size, resulting in I/O errors. The solution was to change md_maxphys to match the maxphys parameter, reboot and recreate the metadevices.

    Copyright:

    © All Rights Reserved
    Download as PDF, TXT or read online from Scribd
    Download as pdf or txt
    of 2

    Document 1544646.1 https://support.oracle.com/epmos/faces/DocumentDisplay?_adf.ctrl-stat...

    Solaris Volume Manger SVM - Metadevices Going Into Maintenance Mode Without Any Related
    Disk/Component I/O Errors (Doc ID 1544646.1)

    APPLIES TO:

    Sun Solaris Volume Manager (SVM) - Version 11.9.0 to 11.11 [Release 11.0]
    Oracle Solaris on SPARC (32-bit)
    Oracle Solaris on SPARC (64-bit)
    Oracle Solaris on x86-64 (64-bit)
    Oracle Solaris on x86 (32-bit)

    SYMPTOMS

    After putting the boot disk under Solaris Volume Manager (SVM) control, md_write and md_read errors are seen during reboot and the
    metadevices are placed into maintenance.

    You may see similar messages as these snippets shown below:

    Mar 11 06:56:07 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d46: write error on /dev/dsk/c0t0d0s5
    Mar 11 06:56:07 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d47: write error on /dev/dsk/c0t1d0s5
    Mar 11 06:56:07 Node2 md_mirror: [ID 104909 kern.warning] WARNING: md: d47: /dev/dsk/c0t1d0s5 needs maintenance
    Mar 11 06:56:07 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d46: write error on /dev/dsk/c0t0d0s5
    Mar 11 06:56:15 Node2 last message repeated 1 time
    Mar 11 11:33:58 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d36: write error on /dev/dsk/c0t1d0s4
    Mar 11 11:54:46 Node2 last message repeated 1 time
    Mar 11 11:57:46 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d36: write error on /dev/dsk/c0t1d0s4
    Mar 11 12:06:38 Node2 last message repeated 1 time
    Mar 11 12:09:26 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d36: write error on /dev/dsk/c0t1d0s4
    Mar 11 12:20:05 Node2 last message repeated 1 time
    Mar 11 20:45:06 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d36: read error on /dev/dsk/c0t1d0s4
    Mar 11 20:45:06 Node2 md_mirror: [ID 104909 kern.warning] WARNING: md: d36: /dev/dsk/c0t1d0s4 needs maintenance
    Mar 11 20:45:06 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d35: read error on /dev/dsk/c0t0d0s4
    Mar 11 20:45:06 Node2 md_mirror: [ID 291867 kern.info] NOTICE: md: d35: Last Erred, retry I/O without B_FAILFAST
    Mar 11 20:45:06 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d35: read error on /dev/dsk/c0t0d0s4
    Mar 11 20:45:06 Node2 md_mirror: [ID 104909 kern.warning] WARNING: md: d35: /dev/dsk/c0t0d0s4 needs maintenance
    Mar 11 20:45:06 Node2 md_mirror: [ID 990438 kern.warning] WARNING: md: d35: /dev/dsk/c0t0d0s4 last erred
    Mar 12 00:21:16 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d47: read error on /dev/dsk/c0t1d0s5
    Mar 12 00:21:16 Node2 md_mirror: [ID 104909 kern.warning] WARNING: md: d47: /dev/dsk/c0t1d0s5 needs maintenance
    Mar 12 00:21:16 Node2 md_mirror: [ID 291867 kern.info] NOTICE: md: d46: Last Erred, retry I/O without B_FAILFAST
    Mar 12 00:21:16 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d46: read error on /dev/dsk/c0t0d0s5
    Mar 12 00:21:16 Node2 last message repeated 1 time
    Mar 12 00:21:16 Node2 md_mirror: [ID 104909 kern.warning] WARNING: md: d46: /dev/dsk/c0t0d0s5 needs maintenance
    Mar 12 00:21:16 Node2 md_mirror: [ID 990438 kern.warning] WARNING: md: d46: /dev/dsk/c0t0d0s5 last erred

    Mar 12 09:24:23 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d35: write error on /dev/dsk/c0t0d0s4
    Mar 12 09:24:25 Node2 ufs: [ID 702911 kern.warning] WARNING: Error writing ufs log
    Mar 12 09:24:25 Node2 ufs: [ID 127457 kern.warning] WARNING: ufs log for / changed state to Error
    Mar 12 09:24:25 Node2 ufs: [ID 616219 kern.warning] WARNING: Please umount(1M) / and run fsck(1M)
    Mar 12 09:24:29 Node2 md_stripe: [ID 641072 kern.warning] WARNING: md: d46: write error on /dev/dsk/c0t0d0s5
    Mar 12 09:24:34 Node2 last message repeated 63 times

    These are simple SVM '1-way mirrors' and as such, SVM will put them into 'last erred' state and allow I/O's to continue. Attempts to recover
    by attaching another submirror fail.

    CHANGES

    Recent Live-Upgrade from 5.9 -> 5.10 using c0t1d0 slice 4 (new_root) and 5 (new_var) for the alternate boot environment (ABE).

    CAUSE

    /etc/system contained:

    set maxphys=8388608
    set md:md_maxphys=838860 <----<<< Maximum I/O size in bytes

    1 of 2 5/12/17, 2:30 PM
    Document 1544646.1 https://support.oracle.com/epmos/faces/DocumentDisplay?_adf.ctrl-stat...

    8388608 / 512 = 16384 = 8k (even multiple of a 512 byte sector/block)

    838860 / 512 = 1638.3984375 = 0.799999237K (uneven multiple)

    By default, md_maxphys is set to maxphys.

    Oracle Solaris Tunable Parameters Reference Manual (page 62 - speaks to the default maxphys)

    This incorrect setting for md_maxphys was causing the SVM I/O problems.

    Another useful general reference for SVM is the Solaris Volume Manager Administration Guide.

    SOLUTION

    Edited /etc/system and changed:

    From
    set md:md_maxphys=838860

    To:
    set md:md_maxphys=8388608
    (matching maxphys).

    and rebooting off physical slices, using metaclear d<#> to delete the current metadevices

    followed by re-creating them and putting the Solaris OS back under SVM control by following the attached references.

    Join Community Discussions:

    To discuss this topic further with Oracle experts and industry peers, we also encourage you to review, join or start a discussion in the My
    Oracle Support Community Solaris Volume Manager.

    REFERENCES

    NOTE:1011940.1 - Solaris Volume Manager (SVM) How To Mirror or encapsulate The Boot / Root Disk Partitions Sparc
    NOTE:1007887.1 - Solaris Volume Manager (SVM) How to Mirror the Boot / Root Disk Partitions for Solaris 10 x86
    NOTE:1389593.1 - Solaris Volume Manager (SVM): Mirroring Root Filesystem In Solaris 11
    NOTE:1005712.1 - Solaris Volume Manager (SVM) How to Recover from Boot Problems by Unmirroring the Root Disk [Video]
    Didn't find what you are looking for?

    2 of 2 5/12/17, 2:30 PM

    You might also like