6 Storage Nodes: No open slots, there are 0 before me

[cojohnson]

@sebastian-roth

I just changed all of my FOG nodes to the “working” branch. In case this changes anything. I am having problems with image replication over FTP.

https://forums.fogproject.org/topic/10891/image-replication-not-working

Anyways, Here is the UEFI Client

#!ipxe
set fog-ip 10.21.0.105
set fog-webroot fog
set boot-url http://${fog-ip}/${fog-webroot}
kernel bzImage32 loglevel=4 initrd=init_32.xz root=/dev/ram0 rw ramdisk_size=127000 web=http://10.21.0.105/fog/ consoleblank=0 rootfstype=ext4 mac=9c:eb:e8:3b:01:0f ftp=10.21.0.105 storage=10.21.0.105:/images/ storageip=10.21.0.105 osid=9 irqpoll chkdsk=0 img=HPStreamG2andG3FY1718 imgType=n imgPartitionType=all imgid=45 imgFormat=0 PIGZ_COMP=-6 hostearly=1 type=down
imgfetch init_32.xz
boot

Here is the Legacy Client

#!ipxe
set fog-ip 10.21.0.105
set fog-webroot fog
set boot-url http://${fog-ip}/${fog-webroot}
kernel bzImage32 loglevel=4 initrd=init_32.xz root=/dev/ram0 rw ramdisk_size=127000 web=http://10.21.0.105/fog/ consoleblank=0 rootfstype=ext4 mac=b4:b5:2f:84:9d:46 ftp=10.21.0.105 storage=10.21.0.105:/images/ storageip=10.21.0.105 osid=5 irqpoll hostname=HP4540n-7297 chkdsk=0 img=HP45xxUniversal imgType=n imgPartitionType=all imgid=46 imgFormat=0 PIGZ_COMP=-6 hostearly=1 type=down
imgfetch init_32.xz
boot

[Sebastian Roth]

@cojohnson Why is OS ID set to different values - one is set to 5 the other is 9…?

[Wayne Workman]

@sebastian-roth likely different images for legacy & uefi. One would be MBR, the other GPT. Just a guess.

[Sebastian Roth]

@cojohnson Please post a picture of your images (so we see “HPStreamG2andG3FY1718” and “HP4540n-7297”) and storages lists here. My guess is that those images are configured to be on different storage groups and therefore the clients talk to different storage nodes…

[Tom Elliott]

@wayne-workman

OS ID 5 = Windows 7
OS ID 9 = Windows 10

The OS ID values have no bearing on “GPT/MBR” partition types, that’s all handled by the image itself. You can have a GPT based Windows 7, and an MBR based Windows 10 image of course. (Or vice-versa, or both the same type).

[Tom Elliott]

Seeing as the image os id’s are different, is it likely the image that’s os id 5 actually only available on the master node? It seems like to from what I can tell.

At the moment both image outputs are showing to 10.21.0.105, is this your master node or an alternate node?

Please remember, the queuing system is not 100% perfect. It attempts to switch to less busy nodes. It has no knowledge of which node will be fastest, of course, it just knows how busy one to another node is. Of course, if the image is unable to be found on other nodes, it can’t do anything with that node. If none of the nodes have the image, the only place it can look is the master node. If the master node has no queue-able slots, it can’t do anything. The first host to try to check in should be “there are 0 before me”, second would be “there is 1 before me”, etc… Of course this is timing dependent too, so it’s possible to have many hosts state that “there are 0 before me”. Normally this doesn’t pose much a problem though and the first to checkin first when a slot becomes available will begin the work. I have added a random timer to try to prevent timing issues as such from occurring however.

[cojohnson]

@sebastian-roth
I do not image Legacy equipment very often… I may not be able to test a solution for a while… Note, I did switch from the developers branch to the working branch just recently. I would like to test it on the “Working” branch

[Sebastian Roth]

@cojohnson Sorry for my late reply. We also need a picture of the Storage settings page.

[Sebastian Roth]

@cojohnson Any news on this?

[cojohnson]

@sebastian-roth Sadly No! I haven’t had the need to image numerous machines at the same time in the middle of the school year. I normally only need to do this over the summer.

[Sebastian Roth]

@cojohnson Well, just let us know when you need further help on this.