diff options
Diffstat (limited to '0 inbox')
| -rw-r--r-- | 0 inbox/2024-08-21.md | 27 | ||||
| -rw-r--r-- | 0 inbox/2024-08-22.md | 25 | ||||
| -rw-r--r-- | 0 inbox/2024-08-23.md | 6 | ||||
| -rw-r--r-- | 0 inbox/2024-08-26.md | 1 | ||||
| -rw-r--r-- | 0 inbox/2024-08-27.md | 11 | ||||
| -rw-r--r-- | 0 inbox/2024-08-28.md | 1 | ||||
| -rw-r--r-- | 0 inbox/2024-08-29.md | 0 | ||||
| -rw-r--r-- | 0 inbox/2024-09-03.md | 1 | ||||
| -rw-r--r-- | 0 inbox/2024-09-10.md | 1 | ||||
| -rw-r--r-- | 0 inbox/2024-09-30.md | 1 | ||||
| -rw-r--r-- | 0 inbox/2024-10-02.md | 6 | ||||
| -rw-r--r-- | 0 inbox/2024-10-04.md | 26 | ||||
| -rw-r--r-- | 0 inbox/2024-10-15.md | 3 | ||||
| -rw-r--r-- | 0 inbox/2024-11-11.md | 3 | ||||
| -rw-r--r-- | 0 inbox/2024-11-12.md | 3 | ||||
| -rw-r--r-- | 0 inbox/antagonist_migratie.md | 27 |
16 files changed, 142 insertions, 0 deletions
diff --git a/0 inbox/2024-08-21.md b/0 inbox/2024-08-21.md new file mode 100644 index 0000000..7408220 --- /dev/null +++ b/0 inbox/2024-08-21.md @@ -0,0 +1,27 @@ +Need to start learning to take notes consistently. My idea now, to kick-off this process, is to open a daily note in obsidian every day at the start of the day in which I will keep crude, quick notes. +Then at the end of the day I shall take some time to go over the notes and refine them. Not yet sure if that refinement will already be the end result or if i should go over refined notes again at maybe the end of each week to see if they really stick. +I should commit notes to git to save them. + +Use taskwarrior to keep track of todo's. +`task <filter> <command> [ <mods> | <args> ]` +commands can be abbreviated as long as they dont become ambigious. +Tasks can be recurring. +See `man task`, specially the CONTEXT section, because i'm always confused about the context, `task context none` to unset the damn thing. +Annotate tasks that are critical with the critical tag, `task <ID> annotate +critical` or `task add +critical blaat` +Upgraded to taskwarrior3. Should set-up a task server to synchronize tasks to so I don't lose all of them if my laptop dies. + +Don't forget to always read man pages. Maybe I should create a detailed note about how to use man pages. + +I will start going to the office full-time from now on. +I should make sure my Obsidian notes and taskwarrior tasks are kept safely in a git repo in case i lose my laptop.l + +Today I learned that we can influence the output of `rabbitmqctl list_queues` , we can append it with keywords of columns that it should show, this also allows us to reorder them to set those pesky names on the end to have nicer formatting. +`rabbitmqctl list_queues messages consumers name` shows messages accumulated in Q and the number of consumers it has and then ends with the name. Of course this is explained in the manpage which I should read more often as noted before. + +I should ask for a diagram of the current systems and take note of it, e.g what is OneCRM / OneHOP / OneHome and all that. + +To upgrade a single package using apt we can use `apt-get install --only-upgrade <packagename>`. + +Need to "relearn" to design stuff and document it. + +Designing structures and a mechanism to return and store info about individual disk backups, we can send that as parent when making incremental backups. Currently we fetch parent reference from checkpoint, which in case of volume's doesn't work of course. It also breaks if, on the testpod, we reset all backups etc but dont do so with all checkpoints.
\ No newline at end of file diff --git a/0 inbox/2024-08-22.md b/0 inbox/2024-08-22.md new file mode 100644 index 0000000..fd1531c --- /dev/null +++ b/0 inbox/2024-08-22.md @@ -0,0 +1,25 @@ +Woke up full of motivation to go to the office. Asked my honey if it's okay to join her by train "is there enough time left?", "yeah" she replied. Then when I come downstairs and she asks me if I can be fast, I tell her maybe it's better you go then bc i worry I will not be able to. Right away she storms off mad and angry... My motivation gone, so I'll work from home.. + +Went to the office anyway by motorcycle, cannot feel like I'm waking up when staying at home anymore. + +**MermaidJS ZenUML** +Optional **Title** +**Participants** can be declared when used or at the start. +**Annotators** give **participants** a special icon, defined at the doc site. +**Aliases** give a short **participant** name a more descriptive name when rendered, e.g: A as Alice <- Alice will be rendered, but the code can use A. +**Messages** between **participants** can be **sync, async, creation or reply**. +**Sync** use A.\<MessageName\> have a block {} for nested messages +**Async** use an arrow A->B: How are you? +**Creation** new keyword new A(optional params) +**Reply** three ways to define: +* Assign var from **Sync** message, e.g: OptionalType a = A.SyncMessage() +* Return A.SyncMessage { return blaat } +* @return or @reply on an **async** message +**Nesting** can be done on both creation and sync messages using blocks {}. +**Comments** can be added using `// comment` and support markdown. +**Loops** can be defined using keywords: **while, for, foreach and loop**. +**Alternative** paths can be sequenced using **if, else** statements. if(blaat) { A.syncMessage(B) } +**Opt** fragments rendered via `opt { ... }`. This can be useful for optional responses or paths. +**Parallel** paths can be defined using `par { ... }`. +**Try/catch/finally** statements are supported as well. +**@Starter(participant)** to make the line start from a **participant**
\ No newline at end of file diff --git a/0 inbox/2024-08-23.md b/0 inbox/2024-08-23.md new file mode 100644 index 0000000..6f41ca2 --- /dev/null +++ b/0 inbox/2024-08-23.md @@ -0,0 +1,6 @@ +Discussed about Automatic Evac (HA). We were unsure about what to do when a local storage goes down. Do we restore the instances from backups to a new node? In that case a DB server might lose important transtactions. So we could also wait until it is fixed. + +Would like Rutger to explain the Freenet BGP load balancing setup in more detail. How do the routers work, what are all those interfaces attached to them? + +A local storage disk backup is stored as a QCOW2 image. An incremental has a relationship to its parent, because they neged to be rebased onto each other. +A volume disk backup is stored as raw data and is accompanied by a file of checksums per block. The extra file is used both to track dirtied blocks for incremental backup.
\ No newline at end of file diff --git a/0 inbox/2024-08-26.md b/0 inbox/2024-08-26.md new file mode 100644 index 0000000..a91a1c9 --- /dev/null +++ b/0 inbox/2024-08-26.md @@ -0,0 +1 @@ +Nothing of note
\ No newline at end of file diff --git a/0 inbox/2024-08-27.md b/0 inbox/2024-08-27.md new file mode 100644 index 0000000..eb75d3f --- /dev/null +++ b/0 inbox/2024-08-27.md @@ -0,0 +1,11 @@ +Changed a few things in the backup service: +* No longer delete all checkpoints when we make a full backup + * TODO: make sure we delete the checkpoint corresponding to a backup when the backup is deleted. + * TODO: Gracefully handle all checkpoints being deleted by openstack +* No longer explicitly specify "incremental" bool, just make it a best effort attempt. + * Still need to work out restores + + +When can we reboot n01.compute vps1-cph8. Its uptime is already 300 days. + +Short call with Jakub. Their priorities have shifted again, the focus on backup will be moved to improving performance of OneHome. They plan to synchronize data from OpenStack to OneHome so they can serve it faster. He asked me whether the customer can influence OpenStack state from within their VM. I told him that probably the only thing they can do to influence OpenStack is shutdown their VM.
\ No newline at end of file diff --git a/0 inbox/2024-08-28.md b/0 inbox/2024-08-28.md new file mode 100644 index 0000000..ac0cac1 --- /dev/null +++ b/0 inbox/2024-08-28.md @@ -0,0 +1 @@ +Had a meeting about onehome performance, they query everything in existence for a customer because of their graphQL workflow. they want to now sync everything in existence to a table so their gQL is fast again.
\ No newline at end of file diff --git a/0 inbox/2024-08-29.md b/0 inbox/2024-08-29.md new file mode 100644 index 0000000..e69de29 --- /dev/null +++ b/0 inbox/2024-08-29.md diff --git a/0 inbox/2024-09-03.md b/0 inbox/2024-09-03.md new file mode 100644 index 0000000..1d8a68e --- /dev/null +++ b/0 inbox/2024-09-03.md @@ -0,0 +1 @@ +one on one Rutger; diff --git a/0 inbox/2024-09-10.md b/0 inbox/2024-09-10.md new file mode 100644 index 0000000..07ff5ba --- /dev/null +++ b/0 inbox/2024-09-10.md @@ -0,0 +1 @@ +EYQBGNF95VORXXVVJ6NKFZF expressVPN
\ No newline at end of file diff --git a/0 inbox/2024-09-30.md b/0 inbox/2024-09-30.md new file mode 100644 index 0000000..fef578b --- /dev/null +++ b/0 inbox/2024-09-30.md @@ -0,0 +1 @@ +- Check freenetvps gobs
\ No newline at end of file diff --git a/0 inbox/2024-10-02.md b/0 inbox/2024-10-02.md new file mode 100644 index 0000000..64c611f --- /dev/null +++ b/0 inbox/2024-10-02.md @@ -0,0 +1,6 @@ +* Fixed n06 testpod, due to installing stuff via pip we had some conflicting packages being used. Pip has been purged and stuff seems to work normally now. +* Investigated DNS issues on testpod lb with Erik. +* Fixed some small issues in the backup service + * Deleted servers couldn't be removed from the periodic backup, that is now fixed. + * Less noisy logging by moving some stuff to debug level. +* Deployed image_id change to prod
\ No newline at end of file diff --git a/0 inbox/2024-10-04.md b/0 inbox/2024-10-04.md new file mode 100644 index 0000000..9c77b37 --- /dev/null +++ b/0 inbox/2024-10-04.md @@ -0,0 +1,26 @@ +What is needed to recreate a deleted server: +* name +* description +* block device mapping +* image properties + * while restoring we need to somehow schedule according to image properties, e.g os_type, or is this already resolved by selecting the correct AZ? + * volumes need to clone volume_image_properties +* port mapping +* availability zone +* flavor +* propertieskjk + + + +port +* name +* description +* allowed_address_pairs +* dns_domain +* dns_name +* extra_dhcp_opts +* fixed_ips +* is_port_security_enabled +* security_group_ids +* qos_policy_id +* network_id
\ No newline at end of file diff --git a/0 inbox/2024-10-15.md b/0 inbox/2024-10-15.md new file mode 100644 index 0000000..8887a01 --- /dev/null +++ b/0 inbox/2024-10-15.md @@ -0,0 +1,3 @@ +Reminder: +* write SOP for fix_allocation +* write SOP for fixing duplicate connector
\ No newline at end of file diff --git a/0 inbox/2024-11-11.md b/0 inbox/2024-11-11.md new file mode 100644 index 0000000..5f3eedb --- /dev/null +++ b/0 inbox/2024-11-11.md @@ -0,0 +1,3 @@ +Sqlalchemy/mysql connections lijken een timeout te hitten +zie: +`https://kibana.one.com/s/openstack/app/discover#/?_g=(filters:!(),refreshInterval:(pause:!t,value:60000),time:(from:now-7d%2Fd,to:now))&_a=(columns:!(),dataSource:(dataViewId:d80a7401-b1fd-43b2-bdfc-79dffb0e8b15,type:dataView),filters:!(),hideChart:!f,interval:auto,query:(language:kuery,query:%22cb3204fa-208f-4b90-95a0-846a9e13852b%22),sort:!(!('@timestamp',desc)))` diff --git a/0 inbox/2024-11-12.md b/0 inbox/2024-11-12.md new file mode 100644 index 0000000..e48d53f --- /dev/null +++ b/0 inbox/2024-11-12.md @@ -0,0 +1,3 @@ +Prometheus: +- scrape config: `roles/prometheus/onecom-prometheus-scrape-config-*.json` +-
\ No newline at end of file diff --git a/0 inbox/antagonist_migratie.md b/0 inbox/antagonist_migratie.md new file mode 100644 index 0000000..c04a00b --- /dev/null +++ b/0 inbox/antagonist_migratie.md @@ -0,0 +1,27 @@ +Migrate VMs + +Not managed by customers, only Antagonist managed +Most of it is backoffice, only few are front facing + +Ovirt (KVM) + + +Possible to change IPs? + +Difficulty: private networking between hosts + databases only ones in the private network + + Should be possible to switch IPs + + +Outsourced routers to DC provider, more difficult to integrate networks + +Most of the things can be redeployed, Rutger notes that maybe all they need is access + +We can use their left-over IP space + +Features +- requires internal network, so want to create networks/routers +- HA/automatic failover? -> Not yet +- Anti-affinity rules? Dont want all the LBs on the same (failing) node + |