Daylight Savings Time Change

[wd40]

The change is coming this weekend here in the US. Is it still a good idea to leave your PC stay on during the change?

Randy

[RayProudfoot]

The change is coming this weekend here in the US. Is it still a good idea to leave your PC stay on during the change?

Randy

Randy,

I left mine on and nothing untoward happened. Just a duplication on WU for the data between 0100-0200 local time. Cumulus was fine.

[steve]

I would say yes. That way the outcome is predictable; you just get a repeated set of timestamped data at the time the clock changes. I've been thinking about this recently, and I'm wondering if the rainfall rate calculation might be affected if it's raining at the time - just something to watch out for.

[wd40]

I will leave it on then and make sure the PC sleep function is disabled.

We are in a recorded number of dry days so I want be able to test the rain counting during the time change. Sure need the rain.

Randy

[wd40]

I left the unit on as suggested. The time change happened without a problem. The only thing that happened was a blob on the graph that was interesting but is no problem at all.

[PaulMy]

I also keep everything running, This morning found that my Cumulus website was not updating after 2:04 am. My Weatherlink page was okay. Cumulus seemed to be working fine and updating WU, CWP and PWS all at the usual update times. However, Cumulus console continuously showed 'Connecting to web site' all the while that the 'Uploading Weather Underground' message appeared to flash in its normal way. In the Cumulus console the Web Update tab was faded and could not do a manual update. I saved the Data directory, and then removed the duplicated data lines of 1:30a and 2:00a in the active Nov file. Shut down Cumulus and at restart all corrected ok , other than the known "Yesterday" data all to 0.00 which will likely correct after midnight (not yet comfortable how to do the edit to correct this immediately). Everything is now good .

p.s. there was no issue in the March DST change.

Paul

[steve]

This morning found that my Cumulus website was not updating after 2:04 am.

I don't think this could have been related to the clock change - unless the change caused a problem at your server. I can't think of any way that the clock change could stop any part of Cumulus working; the only possible effect is the duplicated timestamps in the log file. I suppose it is just about conceivable that the clock change could affect the timers (so ftp could get 'stuck' instead of timing out) but I can't imagine that the Windows timers are dependent on the real time clock.

Note that it's only the timestamps which are duplicated; the data is not and is still valid, so shouldn't really be deleted.

[PaulMy]

Thanks Steve,
I intended my post only to note my experience from the DST change but after your reply decided to check further and found I have been getting a lot of FTP errors: FTP component not responding. This happened this morning at 02:06:09, and then next at 02:18:58 ok until I closed Cumulus at 09:15:45 as mentioned earlier. So likely the 2:06:09 error stopped the website updates. There does not look like there is anything in the error log to indicate why Cumulus was still running ok and uploading successfully to WU, CWOP and PWS but not FTP to website or not being able to manually upload from console. Since my last Cumulus restart on Oct 30 there has been the FTP errors on Oct 31 (once), Nov 2 (once), Nov 3 (8 times), Nov 4 (10 times), Nov 5 (once, and Nov 7 (the 1 time so far). I also checked my Weatherlink log and it had the usual uploads this morning at 12:03, 12:13, 12:23, 12:33, 2:43, 12:53, 1:03, 1:13, 1:23, 1:33, 1:43, 1:53, 2:03, 2:13... with the 1:03 to 1:43 repeated as would be expected from moving back 1 hour, and there are also upload errors in WL long for Nov 1 (once), Nov 3 (once) and Nov 4 (twice) and each of these are at much different times to the Cumulus error on that day. Again thanks for your reply and I will monitor the FTP errors and presume will have to get in touch with Godaddy.com on this.

Steve said:
"Note that it's only the timestamps which are duplicated; the data is not and is still valid, so shouldn't really be deleted."

If I want to add back the removed data lines of 1:30 and 2:00 manually (I have the info in the backed up file) that then will not cause any problems in the future?

Paul

[steve]

So likely the 2:06:09 error stopped the website updates

There's probably some sort of error condition which results in the ftp part of Cumulus 'hanging', when it should be timing out or failing in some other way. I occasionally get problems with my web hosting (more so in the last few weeks!) but Cumulus never fails to carry on again for me when the service comes back up. So it's tricky to diagnose...

If I want to add back the removed data lines of 1:30 and 2:00 manually (I have the info in the backed up file) that then will not cause any problems in the future?

None that I can think of; it does make the graphs look odd, as previously noted, but it means that all the data is there, and (ignoring the timestamps) the data follows the correct sequence. Removing the second set of 0100 to 0200 entries means there is a sudden 'jump' in the data values.

[wd40]

This is not a Cumulus problem but I wondered if anyone else was having these weird time intervals in the Wunderground history for today? I have my interval set at 10 minutes which Cumulus is logging correctly.
Randy

[steve]

weird time intervals in the Wunderground history

Are you using Rapid Fire? If so, the update interval in Cumulus isn't used, and WU decides on the interval in the history (by some strange algorithm that seems to have no logic).