Skip to content

Latest commit

 

History

History
119 lines (78 loc) · 3.04 KB

tips.md

File metadata and controls

119 lines (78 loc) · 3.04 KB
Raw

Debugging mobile phones you don't have

Use SauceLabs (free for open source, for short 11 minutes sessions), and Weinre, see debug-ios-safari-without-iphone.md in this directory.

Docker:

Login as root to a running container, without knowing the root password:

s/d exec -u 0 search bash    # user id 0 = root

# or with docker (not docker-compose):
docker exec -u 0 -it container_name bash

Free up disk space, by deleting old images:

docker image prune  # optionally, add:  --all

How to push images to a local repo, to test in Vagrant: see testing-images-in-vagrant.md.

Redis

Empty the db: (or FLUSHALL to empty all dbs — but Ty uses just one)

$ d/c exec cache redis-cli
> FLUSHDB

ElasticSearch stuff:

If the disk is >= 95% full, ElasticSearch enters read-only mode (or read-delete-only). Once you've freed up disk, you need to tell ElasticSearch about this:

curl -XPUT -H "Content-Type: application/json" \
    http://localhost:9200/_all/_settings \
    -d '{"index.blocks.read_only_allow_delete": null}'

Thereafter ElasticSearch should start working again. Docs: https://www.elastic.co/guide/en/elasticsearch/reference/6.2/disk-allocator.html

List indexes:
http://localhost:9200/_aliases

List everything:
http://localhost:9200/_search?pretty&size=9999

List posts in site 3:
http://localhost:9200/all_english_v1/post/_search?pretty&routing=3&size=9999

Search:
http://localhost:9200/_search?pretty&q=approvedText:zzwwqq2

Status of everything:
http://localhost:9200/_cat?v

Request body search:

$ curl -XGET 'http://localhost:9200/_search' -d '{
    "query" : {
        "term" : { "approvedText" : "something" }
    }
}

Reindex everything: (might take long: minutes/hours/weeks, depending on db size)

curl -XDELETE 'http://localhost:9200/all_english_v1/'
docker-compose restart web app

Nodejs

If you get this error, although not out of disk, when running Gulp:

ENOSPC: no space left on device, watch '/opt/talkyard/server/client/app/page/'

Then, the problem is probably that Gulp needs to watch for changes in really many files, and we need to tell the OS to let Gulp do that, by increasing the number of allowed inode watchers. Append this to /etc/sysctl.conf: (on your host OS, not in a container)

###################################################################
# Nodejs
#
# Nodejs might believe we're out of disk space, unless it can watch really many files:
fs.inotify.max_user_watches=524288

Then reload sysctl.conf, like so: sysctl --system. (Persists across reboots.)

See https://stackoverflow.com/questions/22475849/node-js-error-enospc and npm/npm#1131 (comment)

Downloading backups

You can tell rsync to not download only the current month's backups, e.g.:

rsync -av  \
  --prune-empty-dirs  --include='*2019-12*' --include='*/' --exclude='*' \
  -e 'ssh -i ~/.ssh/id_rsa.pub' \
  root@...:/.../backups/  \
  /home/user/...-backups