[pve-devel] [PATCH] iscsi: fix excessive connection test spam on storage monitoring

[Friedrich Weber]

Hi Stelios,

On 06/07/2025 16:14, Stelios Vailakakis wrote:
> Hi Friedrich,
> 
> Thank you for the warm welcome!
> 
>> This seems like an unrelated change that should be in its own patch,
>> could you please re-send it separately?
> 
> I submitted a separate patch in the format requested at https://
> lists.proxmox.com/pipermail/pve-devel/2025-June/071942.html <https://
> lists.proxmox.com/pipermail/pve-devel/2025-June/071942.html>

Thanks! Just FYI, there is also a more convenient public-inbox instance
for browsing our mailing lists [1].

> 
>> This sounds like #957 [1]. A patch [2] with similar objective was merged
>> [3] and included in Proxmox VE 8.4. Can you check whether your patch is
>> still needed on Proxmox VE 8.4?
> 
> On 8.4.1 with the sole Proxmox node that was hosting the TrueNAS
> instance (proxmox5), I continued to get below errors as well as the
> errors on the TrueNAS syslog after doing a full pveupgrade.
> 
> - proxmox5 does list the hostname instead of the IP address under
> iscsiadm -m session, while 1-4 list the IP address. Forcing this
> behavior on proxmox1 did not cause errors to appear (reverted back to IP
> Address after testing).
> - By utilizing my patch above that includes only hostname support, the
> error continued on proxmox5

So if I understand correctly, after applying only your hostname patch on
top of an up-to-date libpve-storage-perl, you are still seeing the
"connection lost" entries on the iSCSI target? Can you double-check the
version of libpve-storage-perl (e.g. using `pveversion -v | grep
libpve-storage`) on top of which you applied your hostname patch? Could
you post the (anonymized) output of `iscsiadm -m node` and `iscsiadm -m
session` on nodes 1-4 and 5?

> - By reverting back to my patched version which included both fixes
> (excessive connection attempts + hostname support) the errors went away
> from automated checking
> - pvesm status no longer reports the error code 15 when run manually
> - This patch is working from an older version of ISCSIPlugin.pm and I
> haven't checked to see why it's only an issue on this proxmox5 node.
> 
> Jul 06 08:30:55 proxmox5 pvestatd[490256]: command '/usr/bin/iscsiadm --
> mode node --targetname iqn.aaaa-bb.com.redacted.ssd-vm-stor --login'
> failed: exit code 15
> Jul 06 08:30:55 proxmox5 pvestatd[490256]: command '/usr/bin/iscsiadm --
> mode node --targetname iqn.xxxx-yy.com.redacted.vm-stor --login' failed:
> exit code 15
> 
> 
> root at proxmox1:~# pdsh -g pve pveversion 2>/dev/null | dshbak -c
> ----------------
> proxmox[1-2,4-5]
> ----------------
> pve-manager/8.4.1/2a5fa54a8503f96d (running kernel: 6.8.12-11-pve)
> 
> root at proxmox1:~# pdsh -g pve 'pvesm status | grep vm-stor' 2>/dev/null |
> dshbak -c
> ----------------
> proxmox[1-2,4-5]
> ----------------
> lvm-nas-ssd-vm-stor         lvm     active      6442442752      
> 723525632      5718917120   11.23%
> lvm-nas-vm-stor             lvm     active      2684346368    
>  1006669824      1677676544   37.50%
> nas-ssd-vm-stor           iscsi     active               0              
> 0               0    0.00%
> nas-vm-stor               iscsi     active               0              
> 0               0    0.00%
> 
> root at proxmox1:~# pdsh -g pve 'ls -l /usr/share/perl5/PVE/Storage/
> ISCSIPlugin.pm' 2>/dev/null | dshbak -c
> ----------------
> proxmox[1-2,4]
> ----------------
> -rw-r--r-- 1 root root 16849 Apr  7 15:18 /usr/share/perl5/PVE/Storage/
> ISCSIPlugin.pm
> ----------------
> proxmox5
> ----------------
> -rw-r--r-- 1 root root 19767 Jul  6 08:33 /usr/share/perl5/PVE/Storage/
> ISCSIPlugin.pm
> 
> Regards,
> Stelios Vailakakis
> 
> ------------------------------------------------------------------------

Best wishes,

Friedrich

[1] https://lore.proxmox.com/