-
Notifications
You must be signed in to change notification settings - Fork 11
notifications
2023.02.08 FTT: astro-api demo and test servers have been turned off. You now need to install astro-api on your own machine to run this service. If this is not an option, please contact user support at met.no/yr.no to check if you may use their api.met.no (which runs astro-api internally). Thank you all for your understanding and support for all these years.
2022.10.01 FTT: Production has been moved to docker containers. Demo and test servers will be running for some time still, but will at some point be turned off without further notice. Thank you all for your understanding and support.
2022.04.22 FTT: Note change in elevation, azimuth and range -indexes in main branch. This will not affect production. Changes are available on astro-test.met.no, and will be pushed to demo next week.
2022.04.01 FTT: Processing moved to ostack. Servers are now internal, only astro.met.no and astro-test.met.no are external.
2021.11.23 FTT: The process to point astro.met.no and astro-test.met.no aliases to the new servers will start this week. Please note that cosmetic changes to the XML-format may occur.
2021.03.04 FTT: Preparations to migrate servers to Ostack platform will start next week. No changes to operational servers at this time. This is just a warning of major changes (new IP, upgrade to latest astro-api) coming within a few weeks.
2019.10.22 FTT: B-nodes were offline for 5 hours today due to maintenance on the cooling system.
2019.08.28 FTT: The new version of event.pl will add some new reports for some of the events. The new reports will use new report Id's (repId). Please make sure that your software checks repId, even if there is currently only one type of reports for that event.
2019.08.08 FTT: New features are being developed on the test-server. Includes more attributes and json format. Documentation may not be consistent with servers.
2019.06.17 FTT: Production servers were resized from 2 to 4 VCPUs and rebooted. Demo server was not resized.
2019.04.05 FTT: All servers have been upgraded and short.pl appears to run correctly.
2019.04.04 FTT: Servers were unable to process encoded URL requests with the new update, giving problems on api.met.no. All servers were rolled back to previous astro-api version, which causes warning overflow in the log files. We are working hard to correct the encoding problems so that the servers can be upgraded to the new version.
2019.04.01 FTT: Servers are being overloaded, giving production anomalies. We are working hard to solve all issues.
2019.03.25 FTT: New script, short.pl, is being tested on all servers.
2019.03.14 FTT: Bugfix: Fixed step reports do not exceed stop time. All servers have been updated.
2019.02.25 FTT: Additional moon phase information (115) has been added to eventid-110. All servers have been updated.
2019.01.21 FTT: All servers were offline for some 10 minutes at 03:30 UTC, for unknown reasons.
2018.12.11 FTT: Total network failure today. Astro-api servers are offline.
2018.11.16 FTT: HW maintenance in dataroom A this week gave 10 minute downtime on all servers. A similar HW maintenance is planned tuesday 20/11 in dataroom B. Expect a similar downtime.
2018.05.22 FTT: Maximum report length increased from 100 to 1000. No noticable change in performance. All servers have been updated.
2018.05.18 FTT: Planned increase of the maximum report length from 100 to 1000 on 2018.05.22.
2018.05.02 FTT: "Moon shadow angles" added to astro-api. All servers have been upgraded.
2018.04.27 FTT: Maintenance due to water leak in data room A planned 14/5 09-15. A-servers (including astro.met.no) will experience downtime. B-servers should not be affected.
2018.o4.27 FTT: All servers will be updated with date-rounding-error fix. All servers were successfully upgraded.
2018.01.23 FTT: IT warns that planned HW maintenance on 2018.01.18 between 10:00Z and 14Z may result in downtime on all dataroom A servers (servers with _a in their name, in addition to the dem-server).
2018.01.08 FTT: IT warns that planned HW maintenance later this week will result in individual server downtime.
2017.12.23 FTT: The astro.met.no demonstration server and astro-a2 production server were down for 1h45m at around 20:00h due to hardware problems. The other production servers (astro-a1,astro-a3,astro-b1,astro-b2,astro-b3) were not affected.
2017.12.20 FTT: The client legal notice will be standardized to comply with the "met.no" legal notice. This should not affect server performance.
2017.11.23 FTT: All servers will be upgraded with EPHEMERIDE file JPLEPH.1600-2200. Downtime ~10 seconds.
2017.10.23 FTT: All servers are now upgraded with a faster nutation model. Downtime ~10 seconds. New installation processes requests 10x faster than the old installation. 50% of processing time is used for ephemeride interpolation, 30% in SOFA-astronomical-algorithms and 20% in general processing (including 5% in formatting).
2017.10.20 FTT: http://astro.met.no/astro is upgraded with a faster nutation model. Downtime ~10 seconds. Other servers will be upgraded next week.
2017.10.20 FTT: all servers will be upgraded to latest bugfix, which solves to problem of extra grayzone events (for instance sun rise close to maximum solar elevation). Downtime ~10 seconds.
2017.06.26 FTT: all servers will be upgraded to latest bugfixes, including error-reporting. Downtime ~10 seconds.
2017.05.23 FTT: mod_perl now apparently works with astro-api, giving latency of about 1.6ms (which is comparable to api.met.no). The system needs further testing before release 1.0, possibly late this week.
2017.05.12 FTT: Apparently the suggested mod_perl configuration is not stable. After perl overhead and mod_perl was removed, a short astro-api request now takes about 20ms (i.e. 10x slower than at api.met.no).
2017.05.08 FTT: a short request with mod_perl on an "ares VM" now takes ~50ms, opposed to ~150ms without mod_perl. The calculation part is about 1ms, the rest apparently being perl overhead. An clean mod_perl request (without any calculations) is about ~1ms. A direct fortran execution, without mod_perl, is about 6ms. So the fastest option seems to be mod_perl without any perl overhead, for which a short request should take about ~2ms (~1ms clean mod_perl + ~1ms fortran calculations). This would be comparable to the current api.met.no on a dedicated server, which takes about ~1ms. I will rewrite the code so that the interface becomes completely perl-overhead-free next week (with no unecessary processing in the perl code and any communication to the fortran code passed as environment variables, and results passed out using STDOUT).
2017.05.05 FTT: There is an issue with mod_perl apparently not keeping processes alive between sessions, giving a severe performance hit. If the mod_perl issue is not solved, a rewrite of the interface code may become necessary. The interface syntax will not be changed regardless of solution.
2017.05.03 FTT: All self-tests passed: cd astro/src/astro/perl; perl t/10-astroEvent.t
2017.04.27 FTT: astro-api repository has been updated with complete source-code.
2017.04.20 FTT: System is being installed. Anomalies may occur until further notice.