bradg
01-14-2005, 07:55 PM
OK, the first time it happened, I thought it was a status screen refresh "glitch", but I've seen it three times tonite on two different boxes.
We've been having a ton of trouble with the cold tonite, trying to isolate the issues and get OSPF to behave, but that doesn't seem to be the big issue here.
Running Star-OS latest, WRAP boards, CM9 cards. 22dB grids, shots of 100 yards (yes - yards) and about 1/3 mile. "Uplink" units are indoors (above 50F), "relay" unit is outside in the cold. Signal quality on both links in excess of 25 at all times - most of the time well above 30.
Three times tonite I've seen this - twice on the same unit. Routes are a mess because the ring gets broken (associations or connectivity seem to drop for no good reason other than I assume the cold). But I can see the remote unit associated (and sometimes even ping it from the reachable unit, which is odd).
So, in the case of no wireless connectivity, I get into it remotely from the ethernet port and "apply changes" to force a re-association (it seems to work). After connectivity is re-established, no routes come up. I wait several minutes, and finally chain-hop SSH into the unit and go to advanced routing - OSPF shows up as STOPPED, and a second later, shows up as STARTED. Lo-and-behold, about 30 seconds after I touch that menu, routes propagate, and things start to work.
No additional "activate changes", no "save changes", no reboot, nothing. Just touch the advanced routing menu, and OSPF goes from stopped to started, and volia - it's working.
I also notice that if I enter the OSPF configuration terminal and exit, CPU load shoots up dramatically, and then settles down, and sometimes routes then propagate.
Does simply touching that menu magically kick or start what should be an already started routing process (especially after an apply changes)?
Anyone have similar experiences? Lonnie, Tony - thoughts?
I'm *REALLY* trying to figure out where the demons are in the OSPF parts of the Star-OS world (if you haven't been able to tell), and this is yet another one - amongst NMBA setings, neighbors, encryption, and other variables I have yet to make sense of.
Brad
We've been having a ton of trouble with the cold tonite, trying to isolate the issues and get OSPF to behave, but that doesn't seem to be the big issue here.
Running Star-OS latest, WRAP boards, CM9 cards. 22dB grids, shots of 100 yards (yes - yards) and about 1/3 mile. "Uplink" units are indoors (above 50F), "relay" unit is outside in the cold. Signal quality on both links in excess of 25 at all times - most of the time well above 30.
Three times tonite I've seen this - twice on the same unit. Routes are a mess because the ring gets broken (associations or connectivity seem to drop for no good reason other than I assume the cold). But I can see the remote unit associated (and sometimes even ping it from the reachable unit, which is odd).
So, in the case of no wireless connectivity, I get into it remotely from the ethernet port and "apply changes" to force a re-association (it seems to work). After connectivity is re-established, no routes come up. I wait several minutes, and finally chain-hop SSH into the unit and go to advanced routing - OSPF shows up as STOPPED, and a second later, shows up as STARTED. Lo-and-behold, about 30 seconds after I touch that menu, routes propagate, and things start to work.
No additional "activate changes", no "save changes", no reboot, nothing. Just touch the advanced routing menu, and OSPF goes from stopped to started, and volia - it's working.
I also notice that if I enter the OSPF configuration terminal and exit, CPU load shoots up dramatically, and then settles down, and sometimes routes then propagate.
Does simply touching that menu magically kick or start what should be an already started routing process (especially after an apply changes)?
Anyone have similar experiences? Lonnie, Tony - thoughts?
I'm *REALLY* trying to figure out where the demons are in the OSPF parts of the Star-OS world (if you haven't been able to tell), and this is yet another one - amongst NMBA setings, neighbors, encryption, and other variables I have yet to make sense of.
Brad