@hugo said in Replication issue between Master node and Storage node:
I set it to 1 because ONLY for this node, it doesn’ allow me to let it to 0.
Can you explain this? Is the gui limiting you from turning this off or setting to 0? Did you have to do some trick to set to 0?
I see you are using the new gui in 1.5.0RCx series. I want to make sure there isn’t something in the new gui that is limiting its action.
@firej8 if your fog server and pxe booting client are on the same subnet you can use the fog server to eves drop on the pxe booting process. Here is a tutorial: https://forums.fogproject.org/topic/9673/when-dhcp-pxe-booting-process-goes-bad-and-you-have-no-clue
Upload the pcap file to a google drive or dropbox. If you are uncomfortable posting the link to the form use the forum direct message and send the link to either Sebastian or myself and we’ll take a look at the pcap and keep it private. When we have an answer then you are free to tear down the pcap from your google drive.
@donutsrule5 Just for clarity, (choosing ipxe.kpxe isn’t a bad decision) the undionly,kpxe driver is the most common used because it uses the undi driver built into each network card. This driver supports the most network adapters because almost all network drivers have undi built in. With that said some implementations of undi is broken which cause undionly.kpxe to fail. The kernel ipxe.kpxe contains all of the known drivers provided by the iPXE group. So this kernel “should” work on all of the common network interfaces.
@tywyn Well there is another field that identifies the primary mac address for each host. You may find that some computers have multiple mac addresses (ie wired and wireless interfaces). Also we found that Win10 has a unique feature to create random mac addresses (which is a pain for NAC/NAP).
There is a field in the hostMAC table that indicates which mac address is primary, but the query doesn’t work like I expect because the query is testing on the join product. One might think that this would work correctly.
select hostName,hmMAC,hmPrimary from hosts left join hostMAC on hosts.hostID=hostMAC.hmHostID where hostMAC.hmPrimary=1;
But it only returns with hmPrimary == 0
But this one does work correctly
select hostName,hmMAC,hmPrimary from hosts left join hostMAC on hosts.hostID=hostMAC.hmHostID where hostMAC.hmPrimary<>1;
This is really troubling way to get the right answer.
[Edit]
Even switching the query around to search on the mac table returns backwards.
MariaDB [fog]> select hostName,hmMAC,hmPrimary from hostMAC join hosts on hosts.hostID=hostMAC.hmHostID where hostMAC.hmPrimary=1;
+-----------------+-------------------+-----------+
| hostName | hmMAC | hmPrimary |
+-----------------+-------------------+-----------+
| testsys1 | 00:50:56:94:f5:21 | 0 |
| testsys2 | 00:50:56:87:3e:1e | 0 |
| testsys3 | 00:50:56:87:b1:15 | 0 |
| toma | 90:b1:1c:60:d2:14 | 0 |
| testsys5 | 00:0c:29:57:90:3b | 0 |
| testsys9 | 80:19:34:11:13:3f | 0 |
| testsys10 | 90:e2:ba:05:6e:a7 | 0 |
| testsys21 | 90:e2:ba:05:6e:a6 | 0 |
+-----------------+-------------------+-----------+
@firej8 Lets make sure I understand this.
You have 2 different computers plugged into the same cable and both have issues getting dhcp address (that is the message that ipxe is saying in your picture of the client.
If this is true, I’m still leaning towards a spanning tree issue.
Bad cable perhaps?
I have to say managing static IP addresses for 450 hosts is quite a big task. When I first read your post I said to my self “oh my god that’s a big job”. This type of control is usually done with a NAC/NAP tool that will dynamically enable/disable network ports based on the identity of the plugged in computer. Microsoft as a built in NAP service or you can use other tools. I’ve used another FOSS solution called packetfence to control what has access when someone plugs in a device.
To your questions
select hostName,hmMAC from hosts left join hostMAC on hosts.hostID=hostMAC.hmHostID;I would say your idea should work, but make take more effort than you want to put in and is not the right tool to use to restrict access to your network.
Moderator’s note: These may not be related, but cross linking threads: https://forums.fogproject.org/topic/10564/failed-to-get-an-ip-via-dhcp
Moderator’s note: May not be related, but cross linking threads: https://forums.fogproject.org/topic/10599/hp-proliant-dl580-g4
Can you check 2 places.
date.timezone in the php.ini file. ref: http://php.net/manual/en/timezones.phpWell I can tell you what’s wrong, but I’m not sure if the script is working as it should. We may need to get @Wayne-Workman involved to take a peek at the script.
To fix your issue edit the /tftboot/default.ipxe file. That has a hard coded IP address in there. That is probably where the old IP is hiding.
@sebastian-roth I need to fill in a bit of detail here. I helped Hugo setup a partitioned FOG Master server. This is where he has some images in /images and other images in /images2. There is a storage group associated with each images container where the full fog server is the master node in each storage group. From what I gather the storage group that points to /images is replicating just fine to its storage node. But the storage group connected to /images2 is not replicating to its storage node.
@Hugo What we would like to see is a screen shot of “Storage Management -> All Storage Nodes” page. This will tell us if you have the proper grouping right.
We would also like to see your master node configuration for the /images2 container.
@donutsrule5 said in PXE Boot Issues:
Posting here is my last resort.
I can tell you the 7440s pxe boot just fine into FOG. And for the record, you are not quite into FOG just yet. You are at the process of attempting to load the iPXE (separate FOSS project) kernel.
The boot image appears to be undionly.kpxe can you tell me if the target computer is in uefi or bios (legacy) mode?
If your computer is in uefi mode 1. you are sending the wrong kernel name to it you need ipxe.efi 2. you need to ensure that secure boot is disabled.
Are you using the built in network adapter or some kind of usb/docking station nic? This error is unique and I personally haven’t seen it just yet.
@jle Sweet, let us know how it goes.
You wouldn’t have thought it was the root bridge causing the issue. But if the root bridge is on the other side of a slow link, it could cause that type of response. We typically set the core switch with a priority of 4095 to ensure only the core switch is ever the stp root bridge.
@jle I can’t speak for the cisco switches as for the configuration. But it sounds like you did find the root of the issue.
The thing between default spanning tree and the fast protocols is its attempt to find the loop back. You have pessimistic vs optimistic checking. Withe default spanning tree it will listen for up to 27 seconds for a bpdu packet then starts forwarding data with the fast spanning tree protocol it starts forwarding data right away then check for the bpdu packet. FOG / FOS boots so fast, by the time standard spanning tree starts to forward data, FOS has already given up on the booting process.
@firej8 If you plug a pc client into the same jack that went to the DL580, it works? We need to eliminate the “network” as the issue.
The next step after the network has been eliminated is to boot this dl580 using a live linux cd. You can use ubuntu desktop live, or linux mint 18.2 any where you can boot and get to a linux command prompt.
Once booted with a linux live dvd, then we can check to see if plugging in one network port will get an ip address.
Can you place a dumb (unmanaged) switch between the pxe booting computer and your building switch? This (on the surface) appears to be a spanning tree issue, where you don’t have one of the fast STP protocols enabled. The dumb switch will help us test this idea.
[edit] well now that I look at it, it may NOT be a spanning tree issue. I would still do that just so we can rule out a networking issue. But the randomness still makes me think spanning tree.
@ale-com said in Images not deploying after update to 1.3.5:
should I deploy from old server and recapture every image or there’s another solution?
Moderator note: You really should start your own thread because your conditions may not exactly match the OPs post.
But to your question, if you have the capabilities to do so, you will end up with a better solution. FOG 1.4.4 has a newer and faster compression engine called zstd. You can only take full advantage of this newer engine if you capture using it. Don’t forget to set your image format to partclone zstd (the default is still partclone gzip)
@firej8 Well your raid disk may be an issue, but right now lets focus on the network. FWIW on most switches spanning tree (default mode) is enabled by default, because it is a good idea.
How many network adapters are in this server?
And I guess I need to ask, does/has this fog server image other computers? I want to rule out an incorrect setting on the fog server.
I guess let me first say that FOG is intended for Workstation class computers not servers. Typically servers have specialty hardware that requires unique drivers which are not common in the standard linux kernel.
With that said it looks like FOS has the right drivers for the DL580 because it is seeing the network interfaces (Yes!!).
Your network interfaces are being seeing by the iPXE kernel because you are getting into the FOG iPXE menu, it just appears that FOS (customized linux OS that captures and deploys images) is having a hard time connecting to the network.
In one of your pictures, the condition reminds me of a networking issue with Spanning Tree protocol, in that you have the default spanning tree protocol enabled and you are not using one of the fast protocols (RSTP, MSTP, fast-STP, or what ever your switch manufacturer calls it). An easy test for a spanning tree issue is to place a dumb (unmanaged) switch between the pxe booting computer and the building network switch. Then pxe boot.
Lets try this before we move on to the next step.
Oh, one thing I noticed from your second picture is that FOG is sending the 32bit kernels to this DL580, I find that a bit odd since this is surely a pure x64 bit system.
Lastly, the pictures are great because they explain the context of the errors very clearly, well done!
@chris178 There is a right way to do this and an easy way.
Easy route.
chmod 755 *.sh./install.sh