Tags

v1.0.19

- RPM fixes for installers
- documentation accordingly updated

v1.0.18

- builds for [ cal9, cc7 ] X [ open6, ua, uademo ] available, images updated
- renamed the rpms in a more explicit way to distinguish the versions
- tested with cal9.2
- JCOP production

v1.0.17

v1.0.17
- rpm and bin available for 
   - cc7: open6, ua, uademo
   - cal9: open6
- quasar 1.5.18, boost 1.81.0 / 1.75.0 (shared linkage) and cal9, cc7 images updated
- improved PSconnected field, flips to fals if no acquisition for >50secs OR lxi comm down
- HWconnected shows only lxi comm status, unrelieable
- renamed all artifacts for more clarity: OpcUaLxi*<os>.<toolset>
- fixed quasar bug in Server/CMakeLists.txt, added src/QuasarUaTraceHook.cpp # code is empty for open6, ifdefs

v1.0.16

the reconnection behavior will be of "type A": if the network is lost, but power stays ON, and the network comes back. The socket is kept.
-
details
----------
the network socket (tcp/ip) with it's lxi session is NOT reconnected on the PLH side, even when a lxi_disconnect followed by a lxi_connect is made. This stems from the aimTTi firmware. On the server, when doing a lxi_(re-)connect, the new session blocks on a mutex on the tcp/ip lxi comm layer, which means basically that the PLH does not reply. After a long (unknown) time the PLH might drop the session and wait for a new request. The  session-expiry delay is not documented for the PLH. If the opcua-lxi makes a new session request when the PLH is not ready then the server thread for that PLH concerned is stuck forever on the (tcp/ip-) session mutex.
With the new "behavior A", the opcua-lxi server ignores transmission errors and does not close/reopen the socket (lxi disconnect, connect) and just KEEPS the session it has. The communication with the PLH comes back after about 30secs when network is restored. If the socket is not used for a longer time the server's OS might clean it up somehow and the communication would be entirely lost. Then a server and PS restart is needed: this corresponds to a full power cycle anyway. It is recommended to connect the PS power throuugh a PDU in any case so that the power cycling can be done remotely.

v1.0.15

- ENS3221: reconnection issues improved: the ID reply does not always come up correctly after power up
  Protect against that with a loop trying 10 times. Seems to work OK, but did not fix the problem really.
- still, reconnection with PL is shaky, further investigations ("reset")

v1.0.14

- increase certificate validity in ServerConfig.xml from 1 year to 10 years
- some mini-code fixes
- increased reconnection delay from 1000ms to 3000ms to increase chance ( https://its.cern.ch/jira/browse/ENS-32211 ). Also improved logging and a more strict reconnected flag in all cases.

v1.0.13

- CPX400DP also added to the code
- config.xml has to have "none" in the current command, as documented

v1.0.12

v1.0.11

VERSION v1.0.11 FOR JCOP EDMS: https://edms.cern.ch/document/2581500/0.1 - updated the python test for asyncio: python3.9.5 and asyncio may2021 - updated the image for the asyncio tests: gitlab-registry.cern.ch/mludwig/docker:opcua-asyncio.cc7 - checked the ports, OK at 23600

v1.0.10

- solving OPCUA-2272: .open6 should load ServerConfig.xml to avoid port clashes and make it configurable. Also cleanup remaining issues with .open6 build
- solving OPCUA-2124 alongside: quasar 1.5.1
- lxi default port is 23600 as per ServerConfig.xml
- C++17
- no other functional changes, just CI, quasar update and port config

v1.0.9

- increased lxi comm timeout from 1000ms to 5000ms to avoid connection problems, it seems to work more stable

v1.0.8

- increased delay to 1s between LXI commands for the aimTTi PL series
- default for all unknown is 2 seconds

v1.0.7

- improved reporting of connect/disconnect: added field PSConnected
- increased comm delay from 100ms to 150ms (SLOW) since occasionally a few timeouts still pop up with the qmd404
- documentation appended accordingly
- still on quasar 1.4.2

v1.0.6

- tag v1.0.6
- further improved delays for communication: have an asymmetric delay now in the main poling loop which is 800ms + 3*lxi comm delay
- the lxi comm delay is per ps type, for aimTTi PL series it is 100ms
- counter is alive also when trying to reconnect

v1.0.5

cpack finally works:
to install:
sudo rpm -iv ./RPM/OpcUaLxiServer-1.0.5.open6.x86_64.rpm
to uninstall:
sudo rpm -e opcualxiserver-1.0.5-1.x86_64
package goes into /opt on target machine
delay adjustment for specific PS types work also, small bug corrected.

v1.0.4

fixing communication delay (hopefully!)
- updated all builder images
- improved slightly code for comm delays
- comm delays per PS type, hardcoded
- liblxi shared linkage for ua, uademo
- liblxi static for open6

v1.0.3

https://its.cern.ch/jira/browse/OPCUA-2044
added a delay 30ms for all lxi commands to avoid saturation and disconnect from the aim-tti PS. This delay is globally hardcoded right now, but of course it can be refined "per family" later on if needed.

v1.0.2

just a minor fix version, no functionality change. The buildchain has changed somehow, related to compat. This patch fixes the build problems for now. But should go to quasar 1.4.2 anyway soon
- open6 compat at 1.3.0, hardcoded ugly fix in cc7_open6.sh and gitlab_cc7_open6.sh
- some new problems with build chain, two small fixes in scripts and .cmake toolchains
- endpoint message for server/open6
- no idea why ua licensed pipeline actually runs, leave it for now
- testing version to enrico for fwLxiPs development

v1.0.1

Ci corrected, minor fixes in scripts

v1.0.0

first release v1.0.0, based on quasar 1.4.0
- all three toolkits (open6, ua, uademo) tested and ok, downloaded from artifacts
- python asyncio script for ramping and subscription
- documentation updated
- some minor bugs corrected
- status for PS and channel added, fields for individual bits
- CI all stages dockerized and OK
- RPM artifacts available, but not tested