Skip to content

fix(coderd): Detect agent disconnect via inactivity #6528

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 1 commit into from
Mar 13, 2023
Merged

fix(coderd): Detect agent disconnect via inactivity #6528

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
75 changes: 66 additions & 9 deletions coderd/workspaceagents.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,8 +11,10 @@ import (
"net/http"
"net/netip"
"net/url"
"runtime/pprof"
"strconv"
"strings"
"sync"
"time"

"github.com/google/uuid"
Expand Down Expand Up @@ -291,11 +293,12 @@ func (api *API) workspaceAgentPTY(rw http.ResponseWriter, r *http.Request) {
})
return
}
go httpapi.Heartbeat(ctx, conn)

_, wsNetConn := websocketNetConn(ctx, conn, websocket.MessageBinary)
ctx, wsNetConn := websocketNetConn(ctx, conn, websocket.MessageBinary)
defer wsNetConn.Close() // Also closes conn.

go httpapi.Heartbeat(ctx, conn)

agentConn, release, err := api.workspaceAgentCache.Acquire(r, workspaceAgent.ID)
if err != nil {
_ = conn.Close(websocket.StatusInternalError, httpapi.WebsocketCloseSprintf("dial workspace agent: %s", err))
Expand Down Expand Up @@ -606,11 +609,40 @@ func (api *API) workspaceAgentCoordinate(rw http.ResponseWriter, r *http.Request
})
return
}
go httpapi.Heartbeat(ctx, conn)

ctx, wsNetConn := websocketNetConn(ctx, conn, websocket.MessageBinary)
defer wsNetConn.Close()

// We use a custom heartbeat routine here instead of `httpapi.Heartbeat`
// because we want to log the agent's last ping time.
var lastPing time.Time
var pingMu sync.Mutex
go pprof.Do(ctx, pprof.Labels("agent", workspaceAgent.ID.String()), func(ctx context.Context) {
// TODO(mafredri): Is this too frequent? Use separate ping disconnect timeout?
t := time.NewTicker(api.AgentConnectionUpdateFrequency)
defer t.Stop()

for {
select {
case <-t.C:
case <-ctx.Done():
return
}

// We don't need a context that times out here because the ping will
// eventually go through. If the context times out, then other
// websocket read operations will receive an error, obfuscating the
// actual problem.
err := conn.Ping(ctx)
if err != nil {
return
}
pingMu.Lock()
lastPing = time.Now()
pingMu.Unlock()
}
})

firstConnectedAt := workspaceAgent.FirstConnectedAt
if !firstConnectedAt.Valid {
firstConnectedAt = sql.NullTime{
Expand Down Expand Up @@ -654,9 +686,12 @@ func (api *API) workspaceAgentCoordinate(rw http.ResponseWriter, r *http.Request
ctx, cancel := context.WithTimeout(dbauthz.AsSystemRestricted(api.ctx), api.AgentInactiveDisconnectTimeout)
defer cancel()

disconnectedAt = sql.NullTime{
Time: database.Now(),
Valid: true,
// Only update timestamp if the disconnect is new.
if !disconnectedAt.Valid {
disconnectedAt = sql.NullTime{
Time: database.Now(),
Valid: true,
}
}
err := updateConnectionTimes(ctx)
if err != nil {
Expand Down Expand Up @@ -711,15 +746,37 @@ func (api *API) workspaceAgentCoordinate(rw http.ResponseWriter, r *http.Request
return
case <-ticker.C:
}
lastConnectedAt = sql.NullTime{
Time: database.Now(),
Valid: true,

pingMu.Lock()
lastPing := lastPing
pingMu.Unlock()

var connectionStatusChanged bool
if time.Since(lastPing) > api.AgentInactiveDisconnectTimeout {
if !disconnectedAt.Valid {
connectionStatusChanged = true
disconnectedAt = sql.NullTime{
Time: database.Now(),
Valid: true,
}
}
} else {
connectionStatusChanged = disconnectedAt.Valid
// TODO(mafredri): Should we update it here or allow lastConnectedAt to shadow it?
disconnectedAt = sql.NullTime{}
lastConnectedAt = sql.NullTime{
Time: database.Now(),
Valid: true,
}
}
err = updateConnectionTimes(ctx)
if err != nil {
_ = conn.Close(websocket.StatusGoingAway, err.Error())
return
}
if connectionStatusChanged {
api.publishWorkspaceUpdate(ctx, build.WorkspaceID)
}
err := ensureLatestBuild()
if err != nil {
// Disconnect agents that are no longer valid.
Expand Down