-package main
+// Copyright (C) The Lightning Authors. All rights reserved.
+//
+// SPDX-License-Identifier: AGPL-3.0
+
+package lightning
import (
+ "bufio"
"context"
"encoding/json"
"errors"
"net/url"
"os"
"regexp"
- "runtime"
"strings"
"sync"
"time"
"git.arvados.org/arvados.git/sdk/go/arvados"
"git.arvados.org/arvados.git/sdk/go/arvadosclient"
"git.arvados.org/arvados.git/sdk/go/keepclient"
+ "github.com/klauspost/pgzip"
log "github.com/sirupsen/logrus"
"golang.org/x/crypto/blake2b"
"golang.org/x/net/websocket"
"runtime_constraints": rc,
"priority": runner.Priority,
"state": arvados.ContainerRequestStateCommitted,
+ "scheduling_parameters": arvados.SchedulingParameters{
+ Preemptible: false,
+ Partitions: []string{},
+ },
+ "environment": map[string]string{
+ "GOMAXPROCS": fmt.Sprintf("%d", rc.VCPUs),
+ },
},
})
if err != nil {
subscribedUUID := ""
defer func() {
if subscribedUUID != "" {
+ log.Printf("unsubscribe container UUID: %s", subscribedUUID)
client.Unsubscribe(logch, subscribedUUID)
}
}()
fmt.Fprint(os.Stderr, neednewline)
neednewline = ""
if subscribedUUID != "" {
+ log.Printf("unsubscribe container UUID: %s", subscribedUUID)
client.Unsubscribe(logch, subscribedUUID)
}
+ log.Printf("subscribe container UUID: %s", cr.ContainerUUID)
client.Subscribe(logch, cr.ContainerUUID)
subscribedUUID = cr.ContainerUUID
}
if m == nil {
return fmt.Errorf("cannot find uuid in path: %q", *path)
}
- uuid := m[2]
- mnt, ok := runner.Mounts["/mnt/"+uuid]
+ collID := m[2]
+ mnt, ok := runner.Mounts["/mnt/"+collID]
if !ok {
mnt = map[string]interface{}{
"kind": "collection",
- "uuid": uuid,
}
- runner.Mounts["/mnt/"+uuid] = mnt
+ if len(collID) == 27 {
+ mnt["uuid"] = collID
+ } else {
+ mnt["portable_data_hash"] = collID
+ }
+ runner.Mounts["/mnt/"+collID] = mnt
}
- *path = "/mnt/" + uuid + m[3]
+ *path = "/mnt/" + collID + m[3]
}
return nil
}
return coll.UUID, nil
}
+// zopen returns a reader for the given file, using the arvados API
+// instead of arv-mount/fuse where applicable, and transparently
+// decompressing the input if fnm ends with ".gz".
+func zopen(fnm string) (io.ReadCloser, error) {
+ f, err := open(fnm)
+ if err != nil || !strings.HasSuffix(fnm, ".gz") {
+ return f, err
+ }
+ rdr, err := pgzip.NewReader(bufio.NewReaderSize(f, 4*1024*1024))
+ if err != nil {
+ f.Close()
+ return nil, err
+ }
+ return gzipr{rdr, f}, nil
+}
+
+// gzipr wraps a ReadCloser and a Closer, presenting a single Close()
+// method that closes both wrapped objects.
+type gzipr struct {
+ io.ReadCloser
+ io.Closer
+}
+
+func (gr gzipr) Close() error {
+ e1 := gr.ReadCloser.Close()
+ e2 := gr.Closer.Close()
+ if e1 != nil {
+ return e1
+ }
+ return e2
+}
+
var (
arvadosClientFromEnv = arvados.NewClientFromEnv()
+ keepClient *keepclient.KeepClient
siteFS arvados.CustomFileSystem
siteFSMtx sync.Mutex
)
-func open(fnm string) (io.ReadCloser, error) {
+type file interface {
+ io.ReadCloser
+ io.Seeker
+ Readdir(n int) ([]os.FileInfo, error)
+}
+
+func open(fnm string) (file, error) {
if os.Getenv("ARVADOS_API_HOST") == "" {
return os.Open(fnm)
}
if m == nil {
return os.Open(fnm)
}
- uuid := m[2]
- mnt := "/mnt/" + uuid + "/"
- if !strings.HasPrefix(fnm, mnt) {
- return os.Open(fnm)
- }
+ collectionUUID := m[2]
+ collectionPath := m[3]
siteFSMtx.Lock()
defer siteFSMtx.Unlock()
return nil, err
}
ac.Client = arvados.DefaultSecureClient
- kc := keepclient.New(ac)
+ keepClient = keepclient.New(ac)
// Don't use keepclient's default short timeouts.
- kc.HTTPClient = arvados.DefaultSecureClient
- // Guess max concurrent readers, hope to avoid cache
- // thrashing.
- kc.BlockCache = &keepclient.BlockCache{MaxBlocks: runtime.NumCPU() * 3}
- siteFS = arvadosClientFromEnv.SiteFileSystem(kc)
+ keepClient.HTTPClient = arvados.DefaultSecureClient
+ keepClient.BlockCache = &keepclient.BlockCache{MaxBlocks: 4}
+ siteFS = arvadosClientFromEnv.SiteFileSystem(keepClient)
+ } else {
+ keepClient.BlockCache.MaxBlocks += 2
}
- log.Infof("reading %q from %s using Arvados client", fnm[len(mnt):], uuid)
- return siteFS.Open("by_id/" + uuid + "/" + fnm[len(mnt):])
+ log.Infof("reading %q from %s using Arvados client", collectionPath, collectionUUID)
+ f, err := siteFS.Open("by_id/" + collectionUUID + collectionPath)
+ if err != nil {
+ return nil, err
+ }
+ return &reduceCacheOnClose{file: f}, nil
+}
+
+type reduceCacheOnClose struct {
+ file
+ once sync.Once
+}
+
+func (rc *reduceCacheOnClose) Close() error {
+ rc.once.Do(func() { keepClient.BlockCache.MaxBlocks -= 2 })
+ return rc.file.Close()
}