Skip to main content
Ursula_Veksler
New Member
May 20, 2012
Question

HA Active-Passive ' different hard disk status'

  • May 20, 2012
  • 7 replies
  • 11956 views
Hi and thank you in advance! I' ve got 300c HA a-p, worked quite well. Some time ago eventually all logs disappeared from Event log menu, but logs to e-mail was ok, so I didn' t pay mach attention ( my fault :( ) After installing new patch 7 the cluster failed with error - type=event subtype=admin pri=critical vd=root user=" unknown" ui=hasync action=shutdown msg=" User unknown shutdown the device from hasync. The reason is ' slave and master have different hard disk status' " (on master) get sys stat Log hard disk: Need format During the installation process I logged in the slave - everything looked fine, but after all - only master is available and no connection to slave. The problem is that I' m working remotely, so the question is as follow - what would happened if I' ll execute formatlogdisk on master as FG knowledge base recommends? Is there any other way to cleanup disk? Would slave take all traffic to himself and the cluster will get up? Both of them are connected to Internet via same IP.

    7 replies

    ede_pfau
    SuperUser
    SuperUser
    May 21, 2012
    hi, I' d say that you can ' exec formatlogdisk' on the master unit without cluster failover. At the moment it looks like the cluster isn' t running and would not fail over to the slave if needed. What could happen? Formatting will erase the log disk, and might cause a reboot (not sure about this). At least when the master reconnects the slave you will have a reboot of both units to form the cluster. So plan for a short downtime.
    Ursula_Veksler
    New Member
    May 29, 2012
    ede_pfau Thank you once again! I' ll try next weekend :)
    Sylvia
    Explorer
    May 29, 2012
    Hi, we have had the same issue a while ago. Formatting the HD solved the issue. Downtime is unfortunately necessary if your are not on-site (otherwise you could do a manual failover to the second Fortigate, then formatting the first one and let this one boot into the cluster) Regards, Sylvia
    ede_pfau
    SuperUser
    SuperUser
    May 29, 2012
    I' d say a reboot is inevitable because when a cluster forms the interfaces' MAC addresses change to virtual addresses. Cluster operations always are major actions.
    Ursula_Veksler
    New Member
    June 4, 2012
    Hi! The things gone really strange - I' ve execute formatlogdisk, the device rebooted, but with no result. Everything looks the same, cluster was not rebuilt ( In Config -> Advanced: FLASH1 (0MB of 0MB) May it be any hardware problem? # diag hard deviceinfo disk Device S0 14.9 GB ref: 0 USB DISK Pro (USB) [FLASH1] partition 1 14.9 GB ref: 1 label: 7EBFCB682E32C689 [ dev: /dev/sda1 major: 8 minor: 1 free: 0MB mounted: N ] Device S1 14.9 GB ref: 16 USB DISK Pro (USB) [FLASH2] partition 1 256.0 MB ref: n/a label: [ dev: /dev/sdb1 major: 8 minor: 17 free: 225MB mounted: Y ] partition 2 256.0 MB ref: n/a label: [ dev: /dev/sdb2 major: 8 minor: 18 free: 225MB mounted: N ] partition 3 14.4 GB ref: 19 label: 3BD3999B425B66C1 [ dev: /dev/sdb3 major: 8 minor: 19 free: 14329MB mounted: Y ] Total available disks: 2
    ede_pfau
    SuperUser
    SuperUser
    June 4, 2012
    and how does the output of this command look like on the second unit? Is your post from the unit where you can use the disk, or where it fails? You' re not doing these diag commands in vain, Fortinet support will ask for them before issuing a hardware RMA anyway. It might well be a hw failure.
    Ursula_Veksler
    New Member
    June 4, 2012
    From the failed unit, second one is absolutely unavailable remotely... HW fail on a new device after three month of working?! still hope that it is something else... Anyway, I' ve open a ticket, but your answers really more fast and useful :)
    ede_pfau
    SuperUser
    SuperUser
    June 4, 2012
    I' ve overlooked that the cluster hasn' t formed and so you cannot access the 2nd unit via ' exec ha manage' . If this is a really remote location consider a terminal server for remote mgmt.
    Anyway, I' ve open a ticket, but your answers really more fast and useful :)
    thanks, that' s because I' ve got a lot less customers calling in...