Please read the posts in the Announcements section about the current status of Cumulus development now that I have retired

Please read this post before posting

Latest Cumulus release v1.9.4 (build 1099) - Nov 28 2014
Latest Cumulus MX release - v3.0.0 build 3043 Jan 20 2017. See this post for download

Cumulus startup anomalies on my suspect hardware

Discussion specific to Davis weather stations
archae86
Posts: 68
Joined: Sat May 21, 2011 4:46 pm
Weather Station: Davis Wireless Vantage Pro2
Operating System: Windows 7
Location: Albuquerque, NM, USA

Cumulus startup anomalies on my suspect hardware

Postby archae86 » Thu Feb 09, 2017 1:01 am

Background: For several years I have noticed intermittent reception problems logged on my setup, which had a Vantage Pro2 ISS received by an Envoy, linked to a PC by a USB WeatherLink logger, with Cumulus running 99% of the time and WeatherLink started up no less than once every ten days in order to get an intact data log in WL (I use the 5 minute archive interval). In the course of attempting to debug the reception problems, I reconfigured so that now a regular Vantage Pro2 Console is the logger host. After I did so I noticed some additional anomalies, which I initially associated with my "stop Cumulus/download archive to WeatherLink/restart Cumulus" sequence.

On more careful observation the troublesome things seem to be associated with Cumulus startup, regardless of whether a WeatherLink session intervenes. I'm not blaming Cumulus. In fact I've diagnosed (perhaps wrongly) that I own a defective logger, and have ordered new hardware. But I know there are some other Cumulus users who post here who share my habit of brief WL download interruptions, and I'd like you to advise me which of the following possibly non-standard behaviors are present on your hardware when you do a Cumulus exit/WeatherLInk startup, download station, exit/Cumulus startup.

With particular attention to the reception status field at the lower right corner of the Davis console, which can be blinking X at roughly one cycle per four seconds, or R, or L:

I observed that normal reception continued during Cumulus shutdown and WeatherLink startup. During WeatherLInk station download the X stopped blinking (about 30 seconds in my case, as there was 21% of a full archive to download), but resumed as soon as the download ended. On Cumulus restart, quite shortly after startup the console reception status went to R. In three trials, once it got back to normal reception (X blinking) in about three minutes, one in about 7 minutes, and once it timed out and went to L status, from which I recovered it a few hours later by going into the Setup mode. In all cases, the Reception statistics displayed on the diagnostics page, which normally give errors and successes since midnight, were reset to zero. A review of the observed indoor temperature showed an appreciable positive temperature spike at re-connection time. The amount varied, and at maximum was 0.7F reached in less than 25 minutes.

So here is the question set: which of these observations is normal, and which may I consider anomalous behavior of my setup?

1. Console displayed reception status switches to R almost immediately after Cumulus startup?
2. Reception statistics as displayed on the console diagnostic pages reset at Cumulus startup?
3. Prolonged (multi-hour) failure to resume reception, even though the ISS signal strength shows as 35 or so, and the station is detected within two minutes of going into Setup?
4. a readily observable positive spike in reported indoor temperature?

Configuration notes:

The version of Cumulus running is 1.9.4 build 1095 (yes, that is old). The PC is running Windows 7. The ISS was bought as a standard Vantage Pro2 but has solar and UV sensors added later. The console in question is the one originally "kitted" with the ISS in question. Both are a few months under 5 years old. When this console ran in the kitchen for the last 4 years, it and another console run in yet another room routinely reached 99% transmission success scores.
My cumulus generated weather site is at http://pastoll.info/weather

User avatar
PaulMy
Posts: 1548
Joined: Sun Sep 28, 2008 11:54 pm
Weather Station: Davis Vantage Pro 2 w/Envoy USB
Operating System: Windows-8
Location: Komoka, ON Canada
Contact:

Re: Cumulus startup anomalies on my suspect hardware

Postby PaulMy » Thu Feb 09, 2017 1:14 am

My VP2 is 2008 and model 6152 (no solar, no UV, no FARS).
I have the console in the kitchen area as a stand-alone and have not looked at the reception stats for yours, it just keeps on displaying...
I have the Envoy in my computer room with Davis USB logger attached to the weather computer and running Cumulus v.1.9.4 b.1099 at 10 minute logging interval
I shut down Cumulus about the first and 15th of each month to run WL for about 5 minutes.
Then shut down WL and restart Cumulus. Never had any issue on reception, etc. but then have not noticed any reason to check it.

Paul

.

User avatar
steve
Cumulus Author
Posts: 26250
Joined: Mon Jun 02, 2008 6:49 pm
Weather Station: None
Operating System: None
Location: On tour in France
Contact:

Re: Cumulus startup anomalies on my suspect hardware

Postby steve » Thu Feb 09, 2017 6:00 pm

My limited understanding is that the console cannot do more than one thing at a time, and in particular, archive download can prevent it doing anything else, perhaps extending to an inability to keep contact with the ISS. The Davis DLL which Cumulus uses handles the download, so Cumulus has no control over this. The DLL does not handle the download in a very efficient way. One thing you should do is to make sure your Cumulus logger interval is set to the same as the console logger interval.

I believe the DLL does reset the stats as a side effect of reading them.
Steve
-----
Hosting available for Cumulus web sites. See http://sandaysoft.com/forum/viewtopic.php?f=2&t=11876

Please read the posts in the Announcements section about the current status of Cumulus development now that I have retired

User avatar
mcrossley
Posts: 4979
Joined: Thu Jan 07, 2010 9:44 pm
Weather Station: Davis VP2
Operating System: Jessie Lite rPi
Location: Wilmslow, Cheshire, UK
Contact:

Re: Cumulus startup anomalies on my suspect hardware

Postby mcrossley » Fri Feb 10, 2017 8:45 am

The stats get reset by the console after midnight, but they also get reset whenever the time is set via the serial link. Cumulus does this on startup, and in the early hours - configurable via the ini file iirc.

I don't have any problems with the reception stopping in cumulus connect. As Steve said make sure the logger intervals are matched otherwise cumulus will keep the enviy/console/logger busy for some time on reconnect.

BigOkie
Posts: 210
Joined: Tue May 28, 2013 1:06 am
Weather Station: Davis VP2
Operating System: Raspian Jessie
Location: Tulsa, OK

Re: Cumulus startup anomalies on my suspect hardware

Postby BigOkie » Fri Feb 10, 2017 9:29 pm

mcrossley wrote:The stats get reset by the console after midnight, but they also get reset whenever the time is set via the serial link. Cumulus does this on startup, and in the early hours - configurable via the ini file iirc.

I don't have any problems with the reception stopping in cumulus connect. As Steve said make sure the logger intervals are matched otherwise cumulus will keep the enviy/console/logger busy for some time on reconnect.


This is correct..midnight, overnight (4am, and I don't think I've ever seen a way to change that) and on Cumulus restart IIRC.

EDIT: sorry, there is a way on both versions using ClockSettingHour in the INI file. Has to be edited there directly while Cumulus isn't running.

archae86
Posts: 68
Joined: Sat May 21, 2011 4:46 pm
Weather Station: Davis Wireless Vantage Pro2
Operating System: Windows 7
Location: Albuquerque, NM, USA

Re: Cumulus startup anomalies on my suspect hardware

Postby archae86 » Fri Feb 10, 2017 9:58 pm

I've got a new string (Windows 10 PC I built within the last year, new logger, new console, fresh software installs) in the commissioning process. I'm carefully taking a step at a time, so don't have Cumulus installed yet, but regarded the time setting matter, I just performed an experiment.

I had WeatherLink running and used it to adjust the console clock back by two minutes. Within a few seconds the requested time showed up on the console display. Also the console promptly ceased the X transmission heartbeat status and showed R for a little over two minutes before it resumed X heartbeat. Surprised by mcrossley's comment about statistics reset, I checked, and sure enough the reception statistics page was counting up from very recent, not the previous origin.

This aligns with Steve's comment about the console not much being able to do more than one thing at a time in most cases. So it seems multiple types of PC interaction can push the console into R state. (going into console setup, starting Cumulus, setting the time from WeatherLInk, plus likely others)

I'm beginning to think the definition of the problem on my old setup is not what things push it into R, but rather the time it takes it to recover to X, or worse, to fall down to L.

While I can't be sure, I currently imagine many of you have your console pushed into R in cases you don't notice, and that causes you negligible harm as the console recovers to X in a time on the order of no more than a couple of times your logging interval or less.
My cumulus generated weather site is at http://pastoll.info/weather

archae86
Posts: 68
Joined: Sat May 21, 2011 4:46 pm
Weather Station: Davis Wireless Vantage Pro2
Operating System: Windows 7
Location: Albuquerque, NM, USA

Re: Cumulus startup anomalies on my suspect hardware

Postby archae86 » Fri Feb 10, 2017 10:31 pm

BigOkie wrote:there is a way on both versions using ClockSettingHour in the INI file. Has to be edited there directly while Cumulus isn't running.

I searched all the (many) Cumulus .ini files looking for this parameter name. Not there, so obviously I have to add an entry. I was hoping to set it to 5:20 a.m., but the parameter definition at this cumulus.ini documentation page suggests only integers in the range of 0 to 23 are supported. So I quit Cumulus, added a ClockSettingHour=5 entry in the [Station] section, and restarted.

By the way, the other time-slice with greatly elevated reception error rate has been at midnight. Can we confirm or deny that Cumulus also checks for time-discrepancy then, in addition to the time regulated by ClockSettingHour?
My cumulus generated weather site is at http://pastoll.info/weather

BigOkie
Posts: 210
Joined: Tue May 28, 2013 1:06 am
Weather Station: Davis VP2
Operating System: Raspian Jessie
Location: Tulsa, OK

Re: Cumulus startup anomalies on my suspect hardware

Postby BigOkie » Fri Feb 10, 2017 10:38 pm

archae86 wrote:
BigOkie wrote:there is a way on both versions using ClockSettingHour in the INI file. Has to be edited there directly while Cumulus isn't running.

I searched all the (many) Cumulus .ini files looking for this parameter name. Not there, so obviously I have to add an entry. I was hoping to set it to 5:20 a.m., but the parameter definition at this cumulus.ini documentation page suggests only integers in the range of 0 to 23 are supported. So I quit Cumulus, added a ClockSettingHour=5 entry in the [Station] section, and restarted.

By the way, the other time-slice with greatly elevated reception error rate has been at midnight. Can we confirm or deny that Cumulus also checks for time-discrepancy then, in addition to the time regulated by ClockSettingHour?


Cumulus doesn't check for time at midnight; it is in the middle of it's end of day operations. The weather station however resets the receptions stats at that time.

archae86
Posts: 68
Joined: Sat May 21, 2011 4:46 pm
Weather Station: Davis Wireless Vantage Pro2
Operating System: Windows 7
Location: Albuquerque, NM, USA

Re: Cumulus startup anomalies on my suspect hardware

Postby archae86 » Fri Feb 10, 2017 10:58 pm

BigOkie wrote:Cumulus doesn't check for time at midnight; it is in the middle of it's end of day operations. The weather station however resets the receptions stats at that time.

On my current suspicion that my primary problem is erratic recovery of reception after perfectly normal operations causing R status, it would be enough if any of the end of day operations interacted with the console in a way that sometimes triggers R status. Also enough if activity internal to the console had the same effect. I'm not currently eager enough to know the answer to hover over the console at midnight each night for enough nights to get a useful conclusion.
My cumulus generated weather site is at http://pastoll.info/weather

User avatar
mcrossley
Posts: 4979
Joined: Thu Jan 07, 2010 9:44 pm
Weather Station: Davis VP2
Operating System: Jessie Lite rPi
Location: Wilmslow, Cheshire, UK
Contact:

Re: Cumulus startup anomalies on my suspect hardware

Postby mcrossley » Fri Feb 10, 2017 11:07 pm

Looking in my MX diags files, I do see "loop data not received" messages after the time is set, implying the console may be busy doing 'stuff' after the time is set. It doesn't trigger a resync on my station though.

If anything I suspect Cumulus queries the station less at midnight than other times as it is busy doing its own internal day roll-over processing - particularly on a rPi which does not have a very fast processor.

archae86
Posts: 68
Joined: Sat May 21, 2011 4:46 pm
Weather Station: Davis Wireless Vantage Pro2
Operating System: Windows 7
Location: Albuquerque, NM, USA

Re: Cumulus startup anomalies on my suspect hardware

Postby archae86 » Mon Feb 13, 2017 1:40 am

archae86 wrote:So here is the question set: which of these observations is normal, and which may I consider anomalous behavior of my setup?

1. Console displayed reception status switches to R almost immediately after Cumulus startup?
2. Reception statistics as displayed on the console diagnostic pages reset at Cumulus startup?
3. Prolonged (multi-hour) failure to resume reception, even though the ISS signal strength shows as 35 or so, and the station is detected within two minutes of going into Setup?
4. a readily observable positive spike in reported indoor temperature?

I now have over a day of experience running an up-to-date copy of Cumulus1 (1.9.4 build 1099) on a month's-old Windows 10 PC using a newly purchased VP2 console and USB logger.

Based on replication on my new hardware string running up-to-date software, it appears than all of these listed symptoms save for #3 are actually the normal behavior for Cumulus1 interacting with a Davis VP2 console using a USB logger.

I don't have nearly enough experience on the setup to affirm it will prove free of #3, as that was happening at a rate of roughly a probability of 10% on any given day.

I've watched many restarts on both the old and the new string of hardware, and it is right about 6 seconds after Cumulus is launched that I see R status on the console. This coincides to the beginning of the popup-announced download, so seems unrelated to the question of how much or how little data is being downloaded from the logger.

As I mentioned before, even a 20% of logger capacity download to WeatherLink, by contrast, does not trigger R status, though the "heartbeat" blinking X pauses during the active download.

Somewhere I got the impression that the current Cumulus MX beta does not use the Davis DLL for some functions for which Cumulus1 relies on it. If so, I think I should do a trial MX run on my new setup before I do the transition into service and retire the old setup.

One other point: my logger and Cumulus use the same 5 minute sampling interval.
Last edited by archae86 on Sat Feb 18, 2017 11:11 pm, edited 1 time in total.
My cumulus generated weather site is at http://pastoll.info/weather

User avatar
Toxic17
Posts: 543
Joined: Mon May 19, 2014 10:45 pm
Weather Station: Davis Vantage Pro2 Plus
Operating System: Debian 9.1 Stretch
Location: Bangor, NI
Contact:

Re: Cumulus startup anomalies on my suspect hardware

Postby Toxic17 » Mon Feb 13, 2017 7:59 am

archae86 wrote:I now have over a day of experience running an up-to-date copy of Cumulus1 (1.9.4 build 1099) on a month's-old Windows 10 PC using a newly purchased VP2 console and USB logger.


I take it you moved back to cumulus v1.9.4 (1095) since your Website says different, or is this a different setup?
Regards Simon

https://www.conligwx.org
https://twitter.com/conligwx
Davis Vantage Pro2+ - CumulusMX v3.0.0 (build 3043) + Saratoga/PWS
Image

archae86
Posts: 68
Joined: Sat May 21, 2011 4:46 pm
Weather Station: Davis Wireless Vantage Pro2
Operating System: Windows 7
Location: Albuquerque, NM, USA

Re: Cumulus startup anomalies on my suspect hardware

Postby archae86 » Mon Feb 13, 2017 1:24 pm

Toxic17 wrote:I take it you moved back to cumulus v1.9.4 (1095) since your Website says different, or is this a different setup?

I currently have two strings of hardware and software active. My published website remains on the same hardware and software it was running, while my new string I occasionally publish manually. The temporary web address is very like the published one with the substitution of testweather for weather.

The two sites don't look quite the same, as I did not make any modifications yet to the standard templates on the test site, nor did I attempt to import history such as records.
My cumulus generated weather site is at http://pastoll.info/weather

archae86
Posts: 68
Joined: Sat May 21, 2011 4:46 pm
Weather Station: Davis Wireless Vantage Pro2
Operating System: Windows 7
Location: Albuquerque, NM, USA

MX better regarding Davis reception

Postby archae86 » Sat Feb 18, 2017 11:35 pm

Above I mentioned that I had the impression MX might communicate with the Davis console enough differently from Cumulus1 to make a difference in my observations. In particular, I had the impression that Steve had ceased using the Davis DLL for at least some interactions, possibly including logger catchup download at program startup.

Whatever the merits of that thought, I now have actual comparative observations on my newer string of hardware/software. The Cumulus1 on this machine is the current 10.9.4 build 1099. The MX was the current build as of a few days ago (Build 3043), and the WeatherLink is the version which shipped with my recently purchased logger, which is the latest available on the Davis web site (version 6.0.3).

There is a very clear difference at application startup time:

1. C1 always kicks the console into R, MX does not, nor does WL at either startup or logger download
2. C1 always resets the console reception statistics, MX does not, nor does WL.
3. The Stoll9 (new hardware) setup typically recovers faster from R state than does the Stoll6 (older hardware/software which has failed intermittently) setup.
4. On Stoll9, the logged rate of slight reception problems is far higher with C1 running than with MX running, WL running, or none of the three running.

This slight reception problem mentioned as number 4 may be a misinterpretation or red herring. It relies on the WeatherLink plot item "ISS Reception". To observe it, I stop whatever program I am currently evaluating, start WL, download the logger, and make the plot. The variable has a range of 0 to 100%. On the new system, when I am running no weather program at all, or WeatherLink, or Cumulus MX, this variable commonly shows a pure 100% for many hours nonstop. When I run Cumulus1, this variable commonly shows 5-minute samples with less than 100% reception at the rate of one or two per hour, most commonly at 99.1%, but a few a bit worse.

At this very minor level, the observed reception impairment does not, I believe, have a discernible effect on the logged and posted weather data. So it may only be interesting as a possible indication that the Davis hardware somehow finds the MX style of interaction during continuous running a bit less disturbing than the Cumulus1 style of interaction, though the difference is nothing like so pronounced as is the difference during initial logger download at program startup.

As Steve's positioning of MX does not suggest I should be using it for production web site maintenance at this stage in the development and beta-testing, I have no plans to transition my production site. But it is comforting to suppose that if my occasional extended reception problem seen on my old hardware/software string returns to plague my new hardware string running current Cumulus1, there may eventually be relief within the Cumulus world.

I don't have enough testing hours on the new setup to begin to justify a conclusion that it will be free of the extended outage problem. The most hopeful indication is that when I have stopped and restarted Cumulus1 on both the old and new machines simultaneously, the new system has usually recovered from R state faster, sometimes by a minute or two, and already the old machine has stayed in R long enough to drop down to L twice. I suspect this means the new console/logger combination for some reason has a bit easier time receiving my ISS.
My cumulus generated weather site is at http://pastoll.info/weather

User avatar
steve
Cumulus Author
Posts: 26250
Joined: Mon Jun 02, 2008 6:49 pm
Weather Station: None
Operating System: None
Location: On tour in France
Contact:

Re: MX better regarding Davis reception

Postby steve » Sun Feb 19, 2017 12:37 pm

archae86 wrote:As Steve's positioning of MX does not suggest I should be using it for production web site maintenance at this stage in the development and beta-testing, I have no plans to transition my production site.

I have no recommendations one way or the other, but very many people are using MX for their production system. It appears to be as reliable as Cumulus 1, except in a few rare cases where people's systems are not very reliable (Cumulus 1 may handle that slightly better). The main decision is whether you are happy with the different architecture of MX, and whether you can live without those facilities in Cumulus 1 yet to be added to MX. It may be a very very long time before I announce that MX is no longer nominally 'beta'.
Steve
-----
Hosting available for Cumulus web sites. See http://sandaysoft.com/forum/viewtopic.php?f=2&t=11876

Please read the posts in the Announcements section about the current status of Cumulus development now that I have retired


Return to “Davis VP/VP2/Vue”

Who is online

Users browsing this forum: No registered users and 3 guests