Other parts of the site: Search the blog:
Pick a Random Month
- 2024
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- 2023
- December
- November
- October
- September
- August
- June
- May
- April
- March
- February
- January
- 2022
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- 2021
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
- 2020
- December
- November
- October
- September
- August
- July
- June
- May
- April
- March
- February
- January
Full Blog Archive
(This is all the blog posts in cronological order, rebuilt every
night so may be up to 24 hours out of date. It's also a big file
with lots of graphics; please be patient.)
Sun, 01 Jul 2007
To paraphrase Justin “those looking for news of Robert need read no more.”
It appears one of my stumbling blocks in switching over to my backup server running the new OS is that the new OS does not announce all of it’s IP address aliases when it starts up networking. I spent way too many hours trying to figure out why I could only use 1 of the 4 IP addresses I need for my micro-network.
First I was fighting the seemingly random swapping of physical ports with configuration files: sometimes the NIC I was using for my private network would get reconfigured as my public NIC when I rebooted. Next, the networking never seemed to really come up completely configured on reboot.
I got around the first problem with the magic HWADDR= line to match the eth port with a physical MAC address. (And, in fairness, had I not been trying to move configuration files from one machine to another, the install would have done that for me.) I got around the second problem by restarting networking and associated services as the final step in booting up.
That got me to the point of having the first of the 4 addresses working consistently. That’s not sufficient for my needs of course. DNS is provided by two of the missing 3 so I was effectively off the Internet in that configuration. To add confusion to the mix, I had no problems with multiple IP addresses on one card on my internal network. So, of course, I assumed it was a bad card. But it worked fine on the internal network with any of the now three cards in the machine (good thing I figured out the HWADDR magic to make swapping the cards around sane).
The solution (and I’m hoping it holds overnight…) was to arping the upstream router from each of the 4 ip addresses (it really only needed to be done from the 3 missing ones, but I wanted to be consistant). I did that manually and have not baked it into any of the reboot scripts as that seems to hold through a boot. And it held for serveral hours; we’ll see if it makes it over night or if I need to set up some silly cron job for that.
I’m guessing since all my internal server share a single switch with no router involved, arp did not come into play with the internal address — or the other machines knew who to ask for the correct MAC address for those IP addresses.
So, at this point, the temporary server is up and serving the site and I can take a deep breath before starting the upgrade of the “real” server (the in-place upgrade was not going well: I couldn’t boot from the install CD; there may be a known issue about that).
comments...
trackback (explanation)
TrackBack ping me at:
https://www.kayakero.net/news/blosxom/file_11832628325784.trackback
Add a comment...