Weekly Site Update
From ScotGrid
(Difference between revisions)
| Revision as of 11:19, 25 Jun 2012 Mark mitchell (Talk | contribs) Weekly Technical Changes to the Glasgow Scotgrid Cluster ← Go to previous diff |
Revision as of 11:20, 25 Jun 2012 Mark mitchell (Talk | contribs) Weekly Technical Changes to the Glasgow Scotgrid Cluster Go to next diff → |
||
| Line 48: | Line 48: | ||
| ! scope="col" | User Impact | ! scope="col" | User Impact | ||
| ! scope="col" | Progress Status | ! scope="col" | Progress Status | ||
| + | |- | ||
| + | ! scope="row" | Deployment of PerfSonar-PS systems | ||
| + | | 5 Working Days with the week commencing the 25th of June, anticipated|| New network monitoring devices to be added to the Scotgrid Cluster || Low. || | ||
| + | |- | ||
| |- | |- | ||
| ! scope="row" | Scratch disk data cleanup. | ! scope="row" | Scratch disk data cleanup. | ||
Revision as of 11:20, 25 Jun 2012
Pages here are for end users of the ScotGrid systems.
Weekly Technical Changes to the Glasgow Scotgrid Cluster
The purpose of this page is to document the changes that are planned for the Cluster each week.
Please review these and feel free to contact ourselves about these changes.
| Activity | Time Frame | Description | User Impact | Progress Status |
|---|---|---|---|---|
| Migration of WNs to new NFS mount | 5 Working Days with the week commencing the 18th of June, anticipated | The current OS on disk048 is required to be upgraded to SL5 to bring it into line with the rest of the disk estate at GU Scotgrid. As disk048 is the NFS mount for the worker nodes within the cluster a programme to migrate the worker nodes to svr015 is underway first. | Low. WNs are being moved in batches and monitored to minimise impact; timeframe will depend on the speed of the migration | |
| Scientific Linux 5 Upgrade of disk048 | 1 Working Day with the week commencing the 18th of June, anticipated (depends on 1) | Once the WNs have been migrated to svr015, disk048 will be upgraded to SL5 | Low. All other SL4 disk servers have been migrated with low impact in a short timeframe | |
| Partial filesystem draining of disk042 | 2 Working days with the week commencing the 18th of June | As part of the programme to alleviate the hotspot issues seen across the Scotgrid storage recently, the filesystems on disk042 will be partially drained to assess the timescale of this process before performing this action on the rest of the estate. | Low. Data will be redistributed internally | Done |
| Filesystem rebalancing of Scotgrid storage | Week beginning 18th June, possible continuation to following week | Once the test of disk042 is complete, the filesystem rebalancing of the Scotgrid storage will commence. | Low. Data will be redistributed internally | |
| Row heading Channel Bonding of Network Interfaces for Worker Nodes | 1 day week commencing 18th of June | A test involving logically bonding two nic card's on a trial worker node will be conducted this week to test improved line rate throughput | Low. Single Device involved which is not in production. | |
| Row heading Channel bonding of network interfaces in worker nodes #2 | Week commencing 18th of June | Rollout of channel bonding activity in 5) to other WNs. | Low. WNs will be bonded over time once temporarily taken out of production |
| Activity | Time Frame | Description | User Impact | Progress Status |
|---|---|---|---|---|
| Deployment of PerfSonar-PS systems | 5 Working Days with the week commencing the 25th of June, anticipated | New network monitoring devices to be added to the Scotgrid Cluster | Low. | |
| Scratch disk data cleanup. | 5 Working Days with the week commencing the 25th of June, anticipated | As per request from ATLAS dark data on multiple disk servers are being cleaned up. | Low/Medium. Disk Server latency may increase during the clean up. Small possibility that Particle Physics user data if stored in Scratch Disk may be impacted by this clean up operation | |
| Migration of WNs to new NFS mount | 5 Working Days with the week commencing the 18th of June, anticipated | The current OS on disk048 is required to be upgraded to SL5 to bring it into line with the rest of the disk estate at GU Scotgrid. As disk048 is the NFS mount for the worker nodes within the cluster a programme to migrate the worker nodes to svr015 is underway first. | Low. WNs are being moved in batches and monitored to minimise impact; timeframe will depend on the speed of the migration | Ongoing from the week commencing the 18th of June. |
| Scientific Linux 5 Upgrade of disk048 | 1 Working Day with the week commencing the 18th of June, anticipated (depends on 1) | Once the WNs have been migrated to svr015, disk048 will be upgraded to SL5 | Low. All other SL4 disk servers have been migrated with low impact in a short timeframe | Ongoing from the week commencing the 18th of June due to svr015 moves. |
| Partial filesystem draining on multiple disk servers | Week commencing the 25th of June. Several weeks to complete. | As part of the programme to alleviate the hotspot issues seen across the Scotgrid storage recently, the filesystems on multiple disk servers will be partially drained to assess the timescale of this process before performing this action on the rest of the estate. | Low. Data will be redistributed internally | |
| Filesystem rebalancing of Scotgrid storage | Week beginning 25th June, possible continuation to following week | Once the test of disk042 is complete, the filesystem rebalancing of the Scotgrid storage will commence. | Low. Data will be redistributed internally | |
| Row heading Channel Bonding of Network Interfaces for Worker Nodes | 1 day week commencing 18th of June | A test involving logically bonding two nic card's on a trial worker node will be conducted this week to test improved line rate throughput | Low. Single Device involved which is not in production. | Ongoing. |
| Row heading Channel bonding of network interfaces in worker nodes #2 | Week commencing 18th of June | Rollout of channel bonding activity in 5) to other WNs. | Low. WNs will be bonded over time once temporarily taken out of production | Ongoing due to above change being open. |
