- This topic is empty.
20. March 2007 at 9:47 #348
Recently I attempted to migrate our Rapport server to a new box, splitting the database onto a dedicated SQL server. After fiddling with the install table I managed to get the server agents to install and the server talks ok to the database, however I’m having problems with SNMP.
For some reason, SNMP fails to contact the standard service, so device discoveries do not work correctly. The devices are checking in fine with the server but SNMP always fails. The service is running quite happily on the server.
The old server (which the database was created on) is Windows 2000 Server based, whereas the new one is based on Server 2003. I am wondering, considering the changes needed to the database for everything else, whether something in the database which relates to the service on the 2000 box could be preventing it from working correctly on the 2003 machine? Alternatively any other advice would be welcomed.
I already checked service logs and there are no errors from the SNMP service on the server.20. March 2007 at 23:54 #9305
No you need SNMP? I always go into the WDM console after installation and un-tick the SNMP option. SNMP is only used on very old Wyse clients and all new clients use the http agent (HAgent)
If devices do no discover with this agent you have a bigger problem,
-TT21. March 2007 at 14:03 #9309
I’m just wondering how old is very old 🙂
We have a mixture of 3630, 3650, s30, 941G and v90 devices out there. If they all use the HAgent then I’ll just disable SNMP anyway. Could someone confirm if any of those devices need SNMP?
I might add that Wyse “could” have told me this lol but it’s like getting blood from a stone even though we licence around 5000 devices of theirs.21. March 2007 at 21:57 #9310
Yes the 36xx’s would have had the old agent however it should be possible to load the Hagent (the upgrade package is in the agents section in Rapport)
When the devices use the hagent they show up as green and if they use SNMP they are blue. If you look at the properties of the older device it will tell you which agent is loaded on the device.
Apart from SNMP does every thing work, can you push images and packages to all the devices? If this works and you old devices have the been upgraded to the HAgent you are OK. If the old devices were on SNMP still you are back to square one 🙁
-TT22. March 2007 at 1:32 #9312steinlagerMember
- Total Post: 20
- Regular Joe
Im having a very similar problem. My install is workgroup edition 4.5.1 on 2003.
I have a mixture of 3125 3360 and 1200le’s. The 1200le’s checkin perfectly as i was able to update them from 4.2 firmware to 5 using the dhcp option tags and wnos.ini
I have not been able to discover any of the CE terminals at all. If i manually add the http agent to a 3360 or 3125 then it will immediately start working and checkin. Until i manually install the http agent they are completely invisible to WDM. By manually i mean physcially logging onto the winterm and hitting update and pointing to an fTp (wdm) server.
This is a problem as they have hundreds of devices at different locations.
It doesn’t appear to be a routing/switching issue as i am testing this on a local LAN using both directed and udp discovery (have tried different winterms and different hub/switch).
The WDM server is installed on Microsoft Virutal PC… im wondering if this may be stopping the SNMP Broadcast traffic?? Would it be multicast or unicast – would it matter on a VM running the NIC in bridge mode. Has anyone else use WDM in a VMWARE/MS VIRTUAL PC machine to discover winterms using a ‘legacy’ agent and not the Http agent?22. March 2007 at 8:58 #9315
I have used Rapport 4.4.1 in VMware with SNMP without problem, SNMP uses UPD for communication however unless VMware setup is VLAN’ed it should not be a problem.
Has anyone done a packet capture to see what WDM is putting on the wire? I am wondering if this is a general problem with 4.5.1 and SNMP?
-TT22. March 2007 at 11:37 #9317
uhh….what if VMWare IS on a VLAN?22. March 2007 at 11:43 #9318
If it is on its own VLAN (that is different to the clients) you may not be able to image the XPe units. This will not effect your issue with SNMP/discovery or package delivery…22. March 2007 at 11:53 #9319
Ahh I see. Well we don’t do full imaging on the XPE units anyway as the network connection to the remote sites where they are based is too small to think of sending down a 512MB image. We do deliver packages to them, but as you say, that shouldn’t be affected.
It actually appears that the 36xx units do have the http agent rather than the snmp one. In fact the only terminals using the older agent are a bunch of old 3700’s which I’m surprised have lasted this long.22. March 2007 at 12:12 #9320
Wow – 3700’s, could be good paper weights?
Do the devices discover and work correctly on the Hagent? Or can you still not manage them?22. March 2007 at 14:33 #9322
They certainly discovered ok because terminals were checking in, but to be honest we never got as far as testing packaging as we had a timescale and had to rollback to the old server.
I’m in the process of building a fresh test server on 2003 to see if we get the same issues again or not. The only difference will be using a fresh database instead of the existing one. Basically we’re trying to rule out group policy and for that matter everything else. I have a stinky feeling that the database (from the old 2000 server) refers in some way to the SNMP service that breaks it when moving to 2003. If that proves to be the case, I’ll do some more fiddling with the database tables in the copy of the old database I have and see if I can get it to work by removing a few rows.
Which to be honest is exactly what we had to do just to get it to install, as it refers to the old server name several times in one of the tables and has to be manually edited.
Will let you know how it gets on 🙂22. March 2007 at 14:41 #9323
Oh and the 3700’s are the very very last of their kind here. They only weren’t replaced earlier due to an incompatibility with one of our systems and the Powerterm software on the s30’s (which I fixed today) 🙂22. March 2007 at 23:19 #9324
Nice, it would be great if you post the results. In case you did not see these:23. March 2007 at 7:39 #9330
I did see those, however in practice I found that I had to make a couple of specific changes not directly listed there. Without making these changes, certainly on our database, the Rapport 4.4.1 install assumed the server agent was already installed somewhere else.
A list of the additional changes are below and made to the “install” table.
9. Delete the rows with the following entries under the column â€œServerNameâ€
10. Change the entry under the column â€œServerNameâ€ for the row â€œRapport4DBâ€ to the new database server name
- You must be logged in to reply to this topic.