2012-03-20
Replication
Object "A", CN = "CN", MNs = "MN1", "MN2".
A is to be replicated from MN1 to MN2
CN -> MN2.replicate(CNCert, PID="A", node="MN1")
MN2: baseURL = resolve nodeId to baseURL
MN2 -> MN1.getReplica(MN2 Cert, PID="A")
MN1 -> CN.isNodeAuthorized(MN2 subject, PID)
MN1 returns "A"
MN2 -> CN.setReplicationStatus(MN2 Cert, PID="A", "Completed")
2012-03-16 Notes
Updates to Sandbox Needed ?:
Y cn-index
Y mercury
Y cn-metacat
Y cn-os-core
Y cn-portal
Y cn-processdaemon
? cn-rest-service (resolve does not appear to be buggy)
? cn-solr
N cn-version-tool
Preparation for RC release next week
------------------------------------
1. Is the dev environment up-to-date with the latest snapshots as
available in the ubuntu-unstable packages?
- Yes:
- a few changes are needed
- new bouncy castle
- add security provider in /usr/lib/jvm/java-6-sun/jre/lib/security/java.security
- java security extension
- Needs testing, was deployed last night, bouncy castle issue needs to be resolved first
- d1_integration will be used to add content to DEMOs1-4, GMNDEV
- knb war built by hudson unstable is including multiple (4) versions of d1_common, d1_libclient
2. Is the tagging complete for the latest RC?
- remaining to be tagged:
d1_portald1_cn_serviced1_indexerd1_mercury_common have been tagged including build outd1_solr_extensionsd1_portal_servletall of level 7 (cn-buildout), except index, mercury and version tool
3. Does the following build control file have the latest version of all tags in it?
https://repository.dataone.org/software/cicore/trunk/d1_stable_build_control/20120228_control.properties
Yes
4. RC versions of all Debian packages should be available in the
ubuntu-stable repo at http://dev-testing.dataone.org/ubuntu-stable
- Is that repo working out OK? Are the sandbox CNs pointing at this
repo for pulling CN Debian packages?
We haven't pointed the sandbox env here yet: tags first- Yes
5. Is the buildout process working? Is everyone OK with the unstable
and stable build process on Hudson? Any suggestions for improvement?
- to debug any build failures, we need to have the default'String Parameter' updated to the latest tagged release in the Configuration settings of stable software packages
- skye: suggests automating the preparation for tagging
- not critical, but will help in the long run
6. As a general timeline (for discussion):
- Today - we should have Dev environment running with latest snapshots.
We should also have the sandbox installed with the latest RC.
- we'll try!
- Next week (March 5-9) - Advertise availability of sandbox to the
DataONE leadership team plus CCIT for review during next week.
Incorporate suggestions in the unstable (trunk) in preparation for
next (hopefully final) RC.
- confirm that all environments can communicate over each CN required TCP ports
- 22, 443, 389, 5432, 5701, 5702
- Week after next (March 12-16) - build out and advertise next RC for
review. Incorporate changes.
- Week after that (March 22) - Public release on that Thursday
Robert
~~~~~~
* finished up fixing bugs in d1_cn_rest
* built out cn dev environment earlier this morning
* need to add bouncyCastle to dataon-cn-os-core buildout
* need to finish branching and tagging for next release candidate
* need to update all poms after tagging to new snapshot level
* need to test dev environment
* knb buildout in unstable is very unstable
*
Rob
~~~
* consolidated test output for authorization tests into 1 result per node per test
* deployed MNWebTester containing latest libclient_java (1.0.1-RC3)
http://mncheck.test.dataone.org:8080/MNWebTester_v1.0.1-RC3-SNAPSHOT_r7305/
http://mncheck.test.dataone.org:8080
* running tests against DEV environment
* should I create a d1_integration Hudson job for SANDBOX?
Skye
~~~~
* DONE::Finalized tags for index, mercury, cn version tool components.
* DONE::Will be testing mercury3 integration with resolve endpoint and testing metadata view.
* DONE::Build search index and bring mercury3 up on all dev cn.
* Investigating invalid xml character bug discovered during cn dev testing wednesday.
* Added five test cases - still having issue replicating SAX parse exception for invalid xml char
* Going to try catching sax parse exception in index processing and re-retrieve sysMeta object from
hzcast and re-attempt with fresh copy of system metadata.
* Need to discuss with Rob - perhaps can get the test file directly from rob.
Ben
~~~
- DONE changed portal to be environment aware
- DONE will be tagging portal
- 286 objects now, can add many many more. will be getting back to KNB data loading
- PENDING working with LTER folks to stand up an MN
- MN IT tests now passing for Metacat deployment (demo2), including authorization (tier 2) bugs.
- worked on a couple CN replication bugs/details with replication status.
Andy
~~~~
- CLI is working, still testing update and resolve
- will work with Matt on this
- had problems with order of the DN, manually setting the rights_holder can be ambiguous, especially when the openssl tools print DNs backwards from our adopted standard
- Working on data packaging tool.
Roger
~~~~~
- Finished install and working on registration of GMN on mn-stage-orc-1.
- Updated install documentation (was very outdated)
Chris
~~~~~
- DONE
troubleshooting CN Sandbox portal postrgresql session database issue- exception re: no acl in pg_hba.conf for 128.111.36.77, when I think there is
- Roger can't register mn-stage-orc-1 due to lack of registered identity, troubleshooting this - need to chat with Ben
- Was working on performance logging for when we import KNB data to the portal
- Due to multiple CNs/JVMs/threads involved, our approach will be to parse log files
- DONE
Will upgrade the Sandbox env today once the deb packages based on tags are added to the ubuntu-stable channel - Thoughts on split-brain issues: Matt suggested we monitor connections on each CN, and if two are remote CNs are unreachable, shutdown services.
- Suggestion: perhaps the easiest is just to remove firewall allow rules via ufw
- Add the CCIT + Duane to the sandbox as identities
Matt
~~~~
- Gave talk at NEON; received well, will likely become a MN next year; brought up issues of handling v. large data sets (e.g., their AOP effort collects 18TB raw data per day of operation)
- Some discussion of assigning identifiers to continuous sensor streams, and data packaging
Dave
~~~~
March 14th Notes
----------------
Robert
~~~~~~
- Will work to bring Sandbox environment up with RC3 packages.
Skye
~~~~
- Indexing logs and reviewing for errors. found invalid XML error. File not found errors. Maybe rebuild cn-dev.
- Version tool command has been added to debian buildouts.
- Coordinate with Dave what next batch of features to include in next Release.
Ben
~~~
- UNM has resolved network problems.
- Resolved MN tester bugs, metacat may be ready for a tag.
Andy
~~~~
- command line interface to add packaging. discover of package document or determine if one should be built.
- May know some software packages that could help determine network problems between CNs.
Roger
~~~~~
- Replication work. Attempt to complete replication tests today.
Chris
~~~~~
- DONE Sandbox SSL certificate issues w/regard to MNs, installed wrong godaddy intermediate. need to to the bundled intermediate cert
- Sandbox is upgraded to RC3
- Dev environment work: worked with Roger and Ben on replication issues. Ben fixed bugs in metacat replication.
- UNM network bouncing and created havoc in hazelcast. We have to learn how to deal with intermittant CN processing.
- CN dev environment seems stable network wise, probably need to upgrade all packages now.