Its particularly fortunate 
So will the default data dir eventually be ~/.local/share/safe for node AND client?
willie@gagarin:~/.local/share/safe$ tree
.
├── cli
│ └── config.json
└── node
└── record_store
3 directories, 1 file
3 Likes
They should be getting stored in ~/.local/share/safe/node by default.
We have previously been advising setting --root-dir for launching the node, but there’s not really any reason to do that when there is a reasonable default.
1 Like
For me it seems intuitive and useful to have multiple directories here. There could well be different data to be stored for both node and client. Do you think it shouldn’t be like that?
Incidentally, a config file should probably be under ~/.config/safe/cli rather than ~/.local/share/safe/cli. I can look into this also.
1 Like
Yes totally agree
3 Likes
Doing exactly what it says on the tin on a Pi4 so far. 
Uploaded an old safe binaries to
Storing file "safe" of 21140392 bytes..
Successfully stored file "safe" to dc6629c10ff01d564479e10bb5f2fb9bca4343d50c9f41dc0e7d29bab2cc776a
Storing file "safenode" of 19731368 bytes..
Successfully stored file "safenode" to 3c6182f30520334b8dea222ec96af64eb07696d5a0fc6abdf06b9da94e2156fc
and again, I removed the old ~/.safe dir but the safeup remade the ~/.safe dir structure automagically
Writing 100 bytes to "/home/safe/.safe/client/uploaded_files/file_names_2023-06-20_01-11-39"
Its not working for the smaller/older SBCs yet though 
THis is my bananaPi
safe@bpi-safe-02:~$ curl -sSL https://raw.githubusercontent.com/maidsafe/safeup/main/install.sh | sudo bash -s --
[sudo] password for safe:
**************************************
* *
* Installing safeup *
* *
**************************************
Architecture armv7l not supported
4 Likes
Awesome, thanks for running this!
So you’ve actually hit a case where the armv7 binaries are used. I thought that was only on Android devices or such things.
We can support that architecture no problem. Will get that fixed.
4 Likes
Btw, what device is this exactly for armv7? Would I be able to get one of these today? Or have to go on eBay or something? I wouldn’t mind having my own wee suite of these machines for my own testing.
2 Likes
I can give you one to play with if you really want. It was the onboard SATA port that interested me. I thought that might be a perf advantage over USB
3 Likes
oracle arm vps working fine one node online no chunks yet but uploads are going niceley
4 Likes
Noting here due to the prefix of PID: 36112 (not contained in the JSON as is), and the , at the very end per line associated with the metrics component, this broke the parsing on extracting the JSON as is (assumptions from the prior testnet).
[2023-06-20T00:33:50.546220Z DEBUG sn_logging::metrics] PID: 36112 {"physical_cpu_threads":4,"system_cpu_usage_percent":10.377834,"system_total_memory_mb":6895.4355,"system_memory_used_mb":198.00473,"system_memory_usage_percent":2.8715334,"network":{"interface_name":"eth0","bytes_received":13533,"bytes_transmitted":75516,"total_mb_received":666.6563,"total_mb_transmitted":4041.8474},"process":{"cpu_usage_percent":0.20151134,"memory_used_mb":27.275265,"bytes_read":0,"bytes_written":0,"total_mb_read":0.049152,"total_mb_written":3.100672}},
No worries, I quickly adjusted the parsing on my end to account for it on this testnet:
Dashboard is looking a bit sad at the moment… no logs associated with Store Chunks workflows or Request IDs etc yet. Will give it more time, 
.
EDIT: The Unique Chunk Addresses of 1 was a bug in the Grafana panel, when no Chunks were being found. Its fixed now.
7 Likes
The installation went smooth on Ubuntu with the non-sudo installation. I successfully managed to upload and download a file. Great progress to see the downloaded file has its filename.extension when downloaded back. I couldn’t get safenode working and in safe there is no node or option to install it?
5 Likes
Did you try (sudo) safeup node?
1 Like
Btw, I don’t suppose you’re putting that Grafana config in a repo somewhere?
Oh so it’s safeup node, that seems to work yes.
1 Like
It used to be in the old safe client that you could control some of the features of the node from there, but that was problematic for various reasons, so it was removed.
Like @Southside mentioned, if you want a node install, you’d use safeup node.
2 Likes
So the process is
sudo rm -r /usr/local/bin/safe
sudo rm -r ~/.safe
curl -sSL https://raw.githubusercontent.com/maidsafe/safeup/main/install.sh | bash -s -- --client
source ~/.config/safe/env
export SAFE_PEERS=/ip4/68.183.44.129/tcp/38921/p2p/12D3KooWN3P9exk23WW7naW7Kb1rdzZeLpZRHXtF5ntyQtJrsiF7
safe
safe files upload -- file_path_to_upload
safe files download
safeup node
SN_LOG=all safenode --log-dir=/tmp/safenode
1 Like
For this testnet, yes.
But when there’s a new client, you should then be able to do a rolling update. So you would run safe client to get the new version.
I may also implement a safeup update command, that would pull in both updated client and node. However, we also need a command for updating safeup itself, which would need to be distinct. So I’m not sure what the names will be.
1 Like
I would just go with safeup update safeup, safeup update node, safeup update client
Same with npm -i npm or pip install --upgrade pip for upgrading the package manager itself to new version.
1 Like
It will certainly be something along those lines anyway.
1 Like
Also, another thing Josh and I were talking about earlier.
It might be that for some subsequent testnets, considering the fact that the nets now have a duration that can span multiple days, and we release every time we do a merge, that we want to use specific versions for the duration of the testnet.
In this case, you’d do:
safeup client --version <version>
safeup node --version <version>
That’s already implemented.
Case in point: we literally just had a new release 20 minutes ago.
1 Like