I’m updating a lot of the computers in the shack. Basically for more powerful yet power efficient machines. This means there are going to periods when SMS spotting will be unavailable or unreliable over the next few days. This only affects the European server, there’s no change to the North American server.
During the normal activation times for Europeans I’ll have it running on the old server but after 1900Z it will be running on the new one whilst I test things. One of the advantages will be auto recovery after a power failure. The current server cannot be automatically restarted, you have to push and hold a button whilst it chugs into life. The new server can start up itself but I need to change the software to allow for this, hence the need to play with the system. Initial tests show that despite moving from PowerPC to Intel architecture and from a 5 year old Linux kernel to an almost new one the bot worked first time. I’m not sure if that’s because I know what I’m doing after all these years or if I was lucky!
Now to put everything back on the old server…
In reply to MM0FMF:
FB on the upgrade - very good to hear that the transition to another architecture went so smoothly - kudos.
Thanks for making SMS spotting available to those of us who have no wish for expensive phones with big screens and bigger contracts - judging by conversations IRL, people really do appreciate it.
Seasons greetings to all,
In reply to M0MYA:
I’m glad you like it. It was a bit of nonsense that grew legs and now has a life of its own. There are about 170 users now. These things become a bit like a millstone round your neck as I feel obliged to provide a reliable service even though I don’t charge. People spend their hard earned money on the cost of an SMS and I feel I should do my best to ensure they get something in return. I had got the old system to the point that it very rarely crashed and could be left unattended for weeks and weeks. Only power fails stopped it.
Getting the teething problems sorted is always a hard task. I have had 2 failures since trying a live run today. One was some crazy X session problem that locked the server up solid. A power cycle was needed to bring back up and there was lots of nonsense in syslog. Hopefully, I fixed that by ensuring VNC sessions get cleaned up regularly.
Richard’s reported issue is quite amusing. The new machine is TOO fast! Comms between the USB based GSM modem and server are getting corrupted. I need to revisit that code. But for now I peppered the IO routines with 0.1sec sleeps. So by slowing the IO down to what is was like on the old machine it seems to be more stable. Not a valid fix but it will do for now. I mean delays are such an amateur way of fixing a timing issue!
In reply to MM0FMF:
These things become a bit like a millstone round your neck as I feel obliged
to provide a reliable service even though I don’t charge.
I know this feeling from a few different areas of life, let alone the hobby. I level it with myself by reasoning that others are giving back and I benefit from those contributions. It’s why I like to thank folks for their efforts - I know how taxing it can be.
People spend their hard earned money on the cost of an SMS and I feel I
should do my best to ensure they get something in return.
Even on my PAYG tariff (with an add-on) a spot costs me 0.1p. That said, I am very grateful for the reliable service and if I ever meet you in person I’ll gladly put 3000 times that towards a pint!
One was some crazy X session problem that locked the server up solid. A power cycle was needed to
bring back up and there was lots of nonsense in syslog.
Hehe - tell me about it!
But for now I peppered the IO routines with 0.1sec sleeps.