Archive for the ‘Server related’ Category

I waited an extra day just to make sure everything rolled over properly.

A quick summary:

Last week was incredibly rough, ethernia.net expired on Monday despite the fact that I renewed it on time. Contacting the registrar led nowhere, the phone was disconnected, the contact email wasn’t getting relayed. During the night, while I was sick from food poisoning, a friend managed to get someone on the phone, and we all discovered that the one person that ran that registrar company had died about 2 or 3 weeks ago.

Harsh. This is where I got my very first domain name, and I stuck with him for more than 20 years…

I had to engage into a lengthy process to recover the domain name, this part isn’t quite finished but at least it is paid up for the next year and functional.

Everything should be back in order, DNS servers can take a little while to update, but it has now been more than 24 hours and the new information should have propagated everywhere. That means deliveries, updates, and all the other dead realm systems should be back up and running.

Shackle stuff

Finally, a picture with decent lighting.

Work wise, I had to make new bolts for the Morgiana shackles because, after everything was done, the original ones didn’t actually fit. I also started the wrist low-poly, it shouldn’t take very long. The code for the ankle cuffs is mostly functional at this point, but I haven’t found a moment to set up my recording area those chain sounds.

There is a good chance no one will read this post for a week or two…

I’m currently getting an issue related to the primary domain name (ethernia.net), the blog can still be accessed through kyrahabattoir.com for the time being, but quite a few things aren’t going to work right while I get this all solved:

• deliveries
• updates
• documentation wiki

I’d recommend you do your purchases on the marketplace for at least a few days.

Oh, yeah I’m also super sick, for unrelated reasons (but the stress is definitely not helping), and couldn’t get anything done today, we still have some progress from Monday and this weekend, so here is some more in-world tests:

Today I finally had a breakthrough with the skin suit gloves. Once I started exagerating the finger joints, knuckles and creasing the underside it sort of began clicking into place in my head. I’m thinking that mid-biceps, is probably long enough for this type of glove. I haven’t decided yet if I want to add some decorative cuff or trim to the edge, maybe a really simple detail strap or trim. I also have to make matching socks, so I probably shouldn’t overdo it.

There were some weird issues yesterday that made me extremely anxious last night: I STILL do not know what happened but connections to my server backend from SL kept failing, I couldn’t be sure if it was a problem on my end or not and ended up recompiling everything on the server box. I’ve also switched a bunch of the scripts around the dead realm region to use SSL because I was suspecting that the 301 redirects from http to https were the culprit.

It ended up going from dozens of connection failures yesterday to just one today, nearly 12 hours ago. I have no idea if it was my doing or if Linden Lab’s.

I need a new collar for the Valentine dance so I’m making a Vergilius cuff set for VRChat!

I ended up modeling a simple padlock for it today because I don’t have any that isn’t built as a sculpted prim for the TouchBound system. I managed to cram them in the free space of the metal texture UV but they are pretty low-resolution all things considered. Since this version is for VRChat I’ve put it together into a custom skeletal rig that uses ModularAvatar for ease of installation and low object/material count (1 mesh for the entire set and 2 materials).

It’s a bit on the heavy side with 40Ktris for the entire set (collar + 4 sets of cuffs), but I’ll include a more “stripped down” version where the inner side of the straps has been removed, and a couple variants for convenience (collar only, wrists only, etc…)

Once this is sorted I’ll return to the “skin suit” project!

The last 3 hours have been absolute hell as I ended up locked out of the site! I assumed the worst and was ready to purge the entire subdomain… But it turned out that it was a misbehaving outdated plugin.

The feeling of the intense stress coming down was almost… orgasmic.

I’m relocating the documentation for KDC products from kdc.ethernia.net/docs/  to: kdc-docs.ethernia.net. You shouldn’t have to do anything special as the old location will redirect to the new one automatically.

I’m currently doing some server updates and separating the two makes things a little easier on me.

Those past two days have been absolute hell, (and are far from over).

Long story short, I lost my server box and I have been scrambling to restore my (horrible) backups on a new machine. I have a “minimal” setup working, and all the KDC related things such as updater & in-world delivery,… should be working properly now.

But please, do message me if you notice a problem.

Story time

Yesterday I received a mail from my ISP informing me that they had blocked my port 25 due to a spam problem. After checking my mail queue it appears to be filled with HUNDRED of delivery error emails failing to deliver to some remote mail server.

Yeah that’s not supposed to happen…

It took me the entire afternoon after troubleshooting every single part of the server (postfix, dovecot, spamassassin, postgrey,… ) to figure out what was actually going on.

I started believing that my mail server had gone rogue as the offending email would reappear on its own even after the mail queue got cleaned out.

I use fetchmail in a cron job to recover mails from some old addresses that I still occasionally use for confirmation emails that come from server that are unable to deal with postgrey.

As it turns out, one specific mail in that mailbox was being consistently rejected by postfix’s policies, generating a new error email every time and sending it back to fetchmail with a “couldn’t deliver” error message, which caused fetchmail to re-attempt delivery of that email every time.

Conclusion

Sure,  it allowed me to review my mail server configuration and to correct a few oddities but what a waste of time.

Such a feeling of “déjà vu” there.

Our server released the “magic smoke” and completely refused to boot during the weekend. Entire machine lost, data unrecoverable. Thankfully I had a partial backup from the day before so not everything was lost.

Saying that, it’s a partial backup, which means that, while the essential systems are restored ( site/vendor system mainly ) there is still a lot of work to do and a lot of data was lost.

I moved all the kdc/ethernia stuff on a brand new server, DNS entries are a little slow to update it seems. If you’re using things like my console or name2key service, there might be slight discrepancies during the next 48 hours.

If you cannot access the documentation link on the left here, the right URL is http://kdc.ethernia.not/docs/ I appear to have a slight caching issue.

Sorry about the 1-2 hours of malfunction of the delivery system, the KDC server update didn’t go as smoothly as I expected. Everything should be fine for 6 more months :).