Repository navigation
Release v1.3.67
Release v1.3.67
v1.3.67 is an agent update (1.3.67). It follows a review of how the agent measures disks, traffic and memory, checked against df, /proc and the physical links in containers. Agents installed with auto-update move to it within a day. The server and web app are the same as v1.3.66.
Highlights
- Disk: the percentage is now df's Use% (used / (used + available)). On ext4 it reads 2–5 points higher than before, because the space kept for root is no longer counted as free; a full disk now reads 100 % instead of 95 %.
- Disk: ZFS pools are counted, each pool's free space once. A ZFS root used to show only its EFI partition.
- Network: only the machine's own NICs are counted, physical or a VM's virtual NIC. Bridges, bonds, VLANs, veth/tap pairs, tunnels and Proxmox firewall bridges are not. A Proxmox VM's download used to count twice, and as upload too.
- Network: speeds are taken per interface, so a link that disappears (a PPP reconnect) no longer drops the whole reading to zero.
- Memory: in a container, file cache is no longer counted as used; reading a large file used to fill the bar. On a ZFS host, the ARC is not counted as used either.
- A hung network mount no longer delays pushes. It made one push in ten arrive 13 s late, close to the 15 s offline limit.
- The first push after an agent start or update reports measured CPU and traffic, not 0.
- TCPing: a lost probe is stamped with its round instead of 3 s later. At most eight targets are dialled at once.
Compatibility
- Agent 1.3.67 uses the same push format and works with every v1.3 server.
- After the update, ext4 hosts show their disk 2–5 points higher, the same figure as df.
Verified
- Go, with the race detector:
- df tables: a full ext4 disk, a ZFS root, btrfs with NFS, snap and overlay mounts;
- sysfs interface sets;
- meminfo with arcstats;
- cgroup v1 and v2 memory.stat;
- TCPing stamps;
- background disk refresh.
- Builds and go vet on Linux amd64/arm64, macOS amd64/arm64 and Windows amd64.
- In containers, against /proc, df and the physical link:
- CPU within 0.5 points;
- memory exact;
- disk equal to df;
- a Proxmox-like topology reads the uplink's traffic;
- a 512 MiB container reads 4 % with a file in cache (was 53 %);
- a hung df keeps pushes 3 s apart;
- end to end with the v1.3.66 server.
📦 Docker Images
xhh1128/pulse:1.3.67xhh1128/pulse:latest
📥 Downloads
Client binaries
| Platform | File |
|---|---|
| Linux amd64 | probe-client |
| Linux arm64 | probe-client-arm64 |
| macOS Intel (amd64) | probe-client-darwin-amd64 |
| macOS Apple Silicon (arm64) | probe-client-darwin-arm64 |
| Windows amd64 | probe-client.exe |
| Install Script (Linux + macOS) | install.sh |
Server standalone binaries (uploaded by the standalone workflow)
| Platform | File |
|---|---|
| Linux amd64 | pulse-server-standalone-linux-amd64 |
| Linux arm64 | pulse-server-standalone-linux-arm64 |
Quick Install (Linux & macOS):
curl -sSL https://raw.githubusercontent.com/xhhcn/Pulse/main/client/install.sh | sudo bash -s -- --id YOUR_ID --server http://YOUR_SERVER:8008One-click migration to a new host:
PASSWORD="$OLD_ADMIN_PW" \
sudo -E ./scripts/migrate.sh --from https://old.example.com